|
32016
|
1241
|
37
|
2026-05-13T09:03:43.548395+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663023548_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
32014
|
NULL
|
NULL
|
NULL
|
|
32017
|
1240
|
22
|
2026-05-13T09:03:44.166082+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663024166_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
5758433181543905301
|
8632682609560398813
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32018
|
1240
|
23
|
2026-05-13T09:03:44.791336+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663024791_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-3427874874907418502
|
8632612938731629533
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/...
|
32017
|
NULL
|
NULL
|
NULL
|
|
32019
|
1240
|
24
|
2026-05-13T09:03:46.539361+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663026539_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
3690775699117387398
|
9133637656415718295
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32020
|
1241
|
38
|
2026-05-13T09:03:46.498362+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663026498_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-739608836208874291
|
8487586657108372437
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32021
|
1240
|
25
|
2026-05-13T09:03:47.047423+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663027047_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-3516962631480187357
|
8487586657116761047
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export...
|
32019
|
NULL
|
NULL
|
NULL
|
|
32022
|
1241
|
39
|
2026-05-13T09:03:47.059265+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663027059_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false}]...
|
6350431119424943521
|
8632611143443688413
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message...
|
32020
|
NULL
|
NULL
|
NULL
|
|
32023
|
1240
|
26
|
2026-05-13T09:03:47.510462+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663027510_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32024
|
1241
|
40
|
2026-05-13T09:03:47.689060+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663027689_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:05 · Firefox / JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app — Work
⏮ 30s
◀ 10s
⏸ Pause
10s ▶
30s ⏭
10:05
Firefox
iTerm2
Slack
CleanShot X
PhpStorm
Finder
QuickTime Player
Alfred
Raycast
Control Centre
Claude
coreautha
Code
Activity Monitor
Windsurf
Anybox...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:30","depth":13,"bounds":{"left":0.45894283,"top":0.21947326,"width":0.00880984,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":13,"bounds":{"left":0.47573137,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:30","depth":13,"bounds":{"left":0.4921875,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:00","depth":13,"bounds":{"left":0.50880986,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:30","depth":13,"bounds":{"left":0.52526593,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:00","depth":13,"bounds":{"left":0.5415558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:30","depth":13,"bounds":{"left":0.55801195,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:00","depth":13,"bounds":{"left":0.57430184,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:30","depth":13,"bounds":{"left":0.59075797,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:00","depth":13,"bounds":{"left":0.6072141,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:30","depth":13,"bounds":{"left":0.6236702,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:00","depth":13,"bounds":{"left":0.64012635,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:30","depth":13,"bounds":{"left":0.6565825,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:00","depth":13,"bounds":{"left":0.67303854,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:30","depth":13,"bounds":{"left":0.68949467,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:00","depth":13,"bounds":{"left":0.70611703,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:30","depth":13,"bounds":{"left":0.72257316,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:00","depth":13,"bounds":{"left":0.73886305,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:30","depth":13,"bounds":{"left":0.7553192,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:00","depth":13,"bounds":{"left":0.77177525,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:30","depth":13,"bounds":{"left":0.7882314,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:00","depth":13,"bounds":{"left":0.804355,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:30","depth":13,"bounds":{"left":0.82081115,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:00","depth":13,"bounds":{"left":0.83759975,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:30","depth":13,"bounds":{"left":0.8540558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12 May 10:05 · Firefox / JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app — Work","depth":10,"bounds":{"left":0.45628324,"top":0.2661612,"width":0.2237367,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"⏮ 30s","depth":9,"bounds":{"left":0.45694813,"top":0.7186752,"width":0.023936171,"height":0.02434158},"on_screen":true,"help_text":"Ctrl+←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"◀ 10s","depth":9,"bounds":{"left":0.48354387,"top":0.71907425,"width":0.02244016,"height":0.023942538},"on_screen":true,"help_text":"←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"⏸ Pause","depth":9,"bounds":{"left":0.5086436,"top":0.7186752,"width":0.027925532,"height":0.02434158},"on_screen":true,"help_text":"Space","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"10s ▶","depth":9,"bounds":{"left":0.53922874,"top":0.71907425,"width":0.022273935,"height":0.023942538},"on_screen":true,"help_text":"→","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"30s ⏭","depth":9,"bounds":{"left":0.56416225,"top":0.7186752,"width":0.024102394,"height":0.02434158},"on_screen":true,"help_text":"Ctrl+→","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXStaticText","text":"10:05","depth":10,"bounds":{"left":0.85738033,"top":0.7254589,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Firefox","depth":9,"bounds":{"left":0.45761302,"top":0.7609737,"width":0.011801862,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"iTerm2","depth":9,"bounds":{"left":0.47772607,"top":0.7609737,"width":0.011801862,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Slack","depth":9,"bounds":{"left":0.4978391,"top":0.7609737,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CleanShot X","depth":9,"bounds":{"left":0.515625,"top":0.7609737,"width":0.02144282,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"PhpStorm","depth":9,"bounds":{"left":0.545379,"top":0.7609737,"width":0.017287234,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finder","depth":9,"bounds":{"left":0.5709774,"top":0.7609737,"width":0.010970744,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"QuickTime Player","depth":9,"bounds":{"left":0.5902593,"top":0.7609737,"width":0.03025266,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Alfred","depth":9,"bounds":{"left":0.62882316,"top":0.7609737,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Raycast","depth":9,"bounds":{"left":0.6476064,"top":0.7609737,"width":0.013630319,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Control Centre","depth":9,"bounds":{"left":0.66954786,"top":0.7609737,"width":0.025598405,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Claude","depth":9,"bounds":{"left":0.7034575,"top":0.7609737,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"coreautha","depth":9,"bounds":{"left":0.72390294,"top":0.7609737,"width":0.017453458,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Code","depth":9,"bounds":{"left":0.7496675,"top":0.7609737,"width":0.00930851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Activity Monitor","depth":9,"bounds":{"left":0.76728725,"top":0.7609737,"width":0.027426861,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Windsurf","depth":9,"bounds":{"left":0.80302525,"top":0.7609737,"width":0.015791224,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Anybox","depth":9,"bounds":{"left":0.82712764,"top":0.7609737,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-586889167018772289
|
9209072041767611357
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:05 · Firefox / JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app — Work
⏮ 30s
◀ 10s
⏸ Pause
10s ▶
30s ⏭
10:05
Firefox
iTerm2
Slack
CleanShot X
PhpStorm
Finder
QuickTime Player
Alfred
Raycast
Control Centre
Claude
coreautha
Code
Activity Monitor
Windsurf
Anybox...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32025
|
1240
|
27
|
2026-05-13T09:03:48.382047+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663028382_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-6773229448344728745
|
8632612251553655765
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response...
|
32023
|
NULL
|
NULL
|
NULL
|
|
32026
|
1241
|
41
|
2026-05-13T09:03:48.350597+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663028350_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
32024
|
NULL
|
NULL
|
NULL
|
|
32027
|
1240
|
28
|
2026-05-13T09:03:48.895987+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663028895_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32028
|
1240
|
29
|
2026-05-13T09:03:49.882545+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663029882_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-6974090244283365455
|
8487585566196132821
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen...
|
32027
|
NULL
|
NULL
|
NULL
|
|
32029
|
1240
|
30
|
2026-05-13T09:03:50.335701+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663030335_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32030
|
1241
|
42
|
2026-05-13T09:03:50.543116+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663030543_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:10 · PhpStorm / faVsco.js – AutomatedReportGenerated.php
⏮ 30s
◀ 10s
⏸ Pause
10s ▶
30s ⏭
10:10
Firefox
iTerm2
Slack
CleanShot X
PhpStorm
Finder
QuickTime Player
Alfred
Raycast
Control Centre
Claude
coreautha
Code
Activity Monitor
Windsurf
Anybox...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:30","depth":13,"bounds":{"left":0.45894283,"top":0.21947326,"width":0.00880984,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":13,"bounds":{"left":0.47573137,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:30","depth":13,"bounds":{"left":0.4921875,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:00","depth":13,"bounds":{"left":0.50880986,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:30","depth":13,"bounds":{"left":0.52526593,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:00","depth":13,"bounds":{"left":0.5415558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:30","depth":13,"bounds":{"left":0.55801195,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:00","depth":13,"bounds":{"left":0.57430184,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:30","depth":13,"bounds":{"left":0.59075797,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:00","depth":13,"bounds":{"left":0.6072141,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:30","depth":13,"bounds":{"left":0.6236702,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:00","depth":13,"bounds":{"left":0.64012635,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:30","depth":13,"bounds":{"left":0.6565825,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:00","depth":13,"bounds":{"left":0.67303854,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:30","depth":13,"bounds":{"left":0.68949467,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:00","depth":13,"bounds":{"left":0.70611703,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:30","depth":13,"bounds":{"left":0.72257316,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:00","depth":13,"bounds":{"left":0.73886305,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:30","depth":13,"bounds":{"left":0.7553192,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:00","depth":13,"bounds":{"left":0.77177525,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:30","depth":13,"bounds":{"left":0.7882314,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:00","depth":13,"bounds":{"left":0.804355,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:30","depth":13,"bounds":{"left":0.82081115,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:00","depth":13,"bounds":{"left":0.83759975,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:30","depth":13,"bounds":{"left":0.8540558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12 May 10:10 · PhpStorm / faVsco.js – AutomatedReportGenerated.php","depth":10,"bounds":{"left":0.45628324,"top":0.2661612,"width":0.13231383,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"⏮ 30s","depth":9,"bounds":{"left":0.45694813,"top":0.7186752,"width":0.023936171,"height":0.02434158},"on_screen":true,"help_text":"Ctrl+←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"◀ 10s","depth":9,"bounds":{"left":0.48354387,"top":0.71907425,"width":0.02244016,"height":0.023942538},"on_screen":true,"help_text":"←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"⏸ Pause","depth":9,"bounds":{"left":0.5086436,"top":0.7186752,"width":0.027925532,"height":0.02434158},"on_screen":true,"help_text":"Space","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"10s ▶","depth":9,"bounds":{"left":0.53922874,"top":0.71907425,"width":0.022273935,"height":0.023942538},"on_screen":true,"help_text":"→","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"30s ⏭","depth":9,"bounds":{"left":0.56416225,"top":0.7186752,"width":0.024102394,"height":0.02434158},"on_screen":true,"help_text":"Ctrl+→","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXStaticText","text":"10:10","depth":10,"bounds":{"left":0.8580452,"top":0.7254589,"width":0.009142287,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Firefox","depth":9,"bounds":{"left":0.45761302,"top":0.7609737,"width":0.011801862,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"iTerm2","depth":9,"bounds":{"left":0.47772607,"top":0.7609737,"width":0.011801862,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Slack","depth":9,"bounds":{"left":0.4978391,"top":0.7609737,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CleanShot X","depth":9,"bounds":{"left":0.515625,"top":0.7609737,"width":0.02144282,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"PhpStorm","depth":9,"bounds":{"left":0.545379,"top":0.7609737,"width":0.017287234,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finder","depth":9,"bounds":{"left":0.5709774,"top":0.7609737,"width":0.010970744,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"QuickTime Player","depth":9,"bounds":{"left":0.5902593,"top":0.7609737,"width":0.03025266,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Alfred","depth":9,"bounds":{"left":0.62882316,"top":0.7609737,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Raycast","depth":9,"bounds":{"left":0.6476064,"top":0.7609737,"width":0.013630319,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Control Centre","depth":9,"bounds":{"left":0.66954786,"top":0.7609737,"width":0.025598405,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Claude","depth":9,"bounds":{"left":0.7034575,"top":0.7609737,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"coreautha","depth":9,"bounds":{"left":0.72390294,"top":0.7609737,"width":0.017453458,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Code","depth":9,"bounds":{"left":0.7496675,"top":0.7609737,"width":0.00930851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Activity Monitor","depth":9,"bounds":{"left":0.76728725,"top":0.7609737,"width":0.027426861,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Windsurf","depth":9,"bounds":{"left":0.80302525,"top":0.7609737,"width":0.015791224,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Anybox","depth":9,"bounds":{"left":0.82712764,"top":0.7609737,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-3120079691316732821
|
8632611289464187869
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:10 · PhpStorm / faVsco.js – AutomatedReportGenerated.php
⏮ 30s
◀ 10s
⏸ Pause
10s ▶
30s ⏭
10:10
Firefox
iTerm2
Slack
CleanShot X
PhpStorm
Finder
QuickTime Player
Alfred
Raycast
Control Centre
Claude
coreautha
Code
Activity Monitor
Windsurf
Anybox...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32031
|
1241
|
43
|
2026-05-13T09:03:51.158778+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663031158_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:10 · Slack / Petko Kashinski (DM) - Jiminny Inc - 5 new items - Slack
⏮ 30s
◀ 10s
⏸ Pause
10s ▶
30s ⏭
10:10
Firefox
iTerm2
Slack
CleanShot X
PhpStorm
Finder
QuickTime Player
Alfred
Raycast
Control Centre
Claude
coreautha
Code
Activity Monitor
Windsurf
Anybox...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:30","depth":13,"bounds":{"left":0.45894283,"top":0.21947326,"width":0.00880984,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":13,"bounds":{"left":0.47573137,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:30","depth":13,"bounds":{"left":0.4921875,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:00","depth":13,"bounds":{"left":0.50880986,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:30","depth":13,"bounds":{"left":0.52526593,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:00","depth":13,"bounds":{"left":0.5415558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:30","depth":13,"bounds":{"left":0.55801195,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:00","depth":13,"bounds":{"left":0.57430184,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:30","depth":13,"bounds":{"left":0.59075797,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:00","depth":13,"bounds":{"left":0.6072141,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:30","depth":13,"bounds":{"left":0.6236702,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:00","depth":13,"bounds":{"left":0.64012635,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:30","depth":13,"bounds":{"left":0.6565825,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:00","depth":13,"bounds":{"left":0.67303854,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:30","depth":13,"bounds":{"left":0.68949467,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:00","depth":13,"bounds":{"left":0.70611703,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:30","depth":13,"bounds":{"left":0.72257316,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:00","depth":13,"bounds":{"left":0.73886305,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:30","depth":13,"bounds":{"left":0.7553192,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:00","depth":13,"bounds":{"left":0.77177525,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:30","depth":13,"bounds":{"left":0.7882314,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:00","depth":13,"bounds":{"left":0.804355,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:30","depth":13,"bounds":{"left":0.82081115,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:00","depth":13,"bounds":{"left":0.83759975,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:30","depth":13,"bounds":{"left":0.8540558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12 May 10:10 · Slack / Petko Kashinski (DM) - Jiminny Inc - 5 new items - Slack","depth":10,"bounds":{"left":0.45628324,"top":0.2661612,"width":0.14760639,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"⏮ 30s","depth":9,"bounds":{"left":0.45694813,"top":0.7186752,"width":0.023936171,"height":0.02434158},"on_screen":true,"help_text":"Ctrl+←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"◀ 10s","depth":9,"bounds":{"left":0.48354387,"top":0.71907425,"width":0.02244016,"height":0.023942538},"on_screen":true,"help_text":"←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"⏸ Pause","depth":9,"bounds":{"left":0.5086436,"top":0.7186752,"width":0.027925532,"height":0.02434158},"on_screen":true,"help_text":"Space","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"10s ▶","depth":9,"bounds":{"left":0.53922874,"top":0.71907425,"width":0.022273935,"height":0.023942538},"on_screen":true,"help_text":"→","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"30s ⏭","depth":9,"bounds":{"left":0.56416225,"top":0.7186752,"width":0.024102394,"height":0.02434158},"on_screen":true,"help_text":"Ctrl+→","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXStaticText","text":"10:10","depth":10,"bounds":{"left":0.8580452,"top":0.7254589,"width":0.009142287,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Firefox","depth":9,"bounds":{"left":0.45761302,"top":0.7609737,"width":0.011801862,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"iTerm2","depth":9,"bounds":{"left":0.47772607,"top":0.7609737,"width":0.011801862,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Slack","depth":9,"bounds":{"left":0.4978391,"top":0.7609737,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CleanShot X","depth":9,"bounds":{"left":0.515625,"top":0.7609737,"width":0.02144282,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"PhpStorm","depth":9,"bounds":{"left":0.545379,"top":0.7609737,"width":0.017287234,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finder","depth":9,"bounds":{"left":0.5709774,"top":0.7609737,"width":0.010970744,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"QuickTime Player","depth":9,"bounds":{"left":0.5902593,"top":0.7609737,"width":0.03025266,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Alfred","depth":9,"bounds":{"left":0.62882316,"top":0.7609737,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Raycast","depth":9,"bounds":{"left":0.6476064,"top":0.7609737,"width":0.013630319,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Control Centre","depth":9,"bounds":{"left":0.66954786,"top":0.7609737,"width":0.025598405,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Claude","depth":9,"bounds":{"left":0.7034575,"top":0.7609737,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"coreautha","depth":9,"bounds":{"left":0.72390294,"top":0.7609737,"width":0.017453458,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Code","depth":9,"bounds":{"left":0.7496675,"top":0.7609737,"width":0.00930851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Activity Monitor","depth":9,"bounds":{"left":0.76728725,"top":0.7609737,"width":0.027426861,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Windsurf","depth":9,"bounds":{"left":0.80302525,"top":0.7609737,"width":0.015791224,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Anybox","depth":9,"bounds":{"left":0.82712764,"top":0.7609737,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-2780111532877277241
|
8632611293759155165
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:10 · Slack / Petko Kashinski (DM) - Jiminny Inc - 5 new items - Slack
⏮ 30s
◀ 10s
⏸ Pause
10s ▶
30s ⏭
10:10
Firefox
iTerm2
Slack
CleanShot X
PhpStorm
Finder
QuickTime Player
Alfred
Raycast
Control Centre
Claude
coreautha
Code
Activity Monitor
Windsurf
Anybox...
|
32030
|
NULL
|
NULL
|
NULL
|
|
32032
|
1240
|
31
|
2026-05-13T09:03:52.972047+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663032972_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-1854564471249859852
|
8632612240816221149
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)...
|
32029
|
NULL
|
NULL
|
NULL
|
|
32033
|
1241
|
44
|
2026-05-13T09:03:52.964664+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663032964_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32034
|
1240
|
32
|
2026-05-13T09:03:53.426+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663033426_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32035
|
1241
|
45
|
2026-05-13T09:03:54.191879+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663034191_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:12 · Firefox / planhat - Google Search — Work
⏮ 30s
◀ 10s
⏸ Pause
10s ▶...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:30","depth":13,"bounds":{"left":0.45894283,"top":0.21947326,"width":0.00880984,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":13,"bounds":{"left":0.47573137,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:30","depth":13,"bounds":{"left":0.4921875,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:00","depth":13,"bounds":{"left":0.50880986,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:30","depth":13,"bounds":{"left":0.52526593,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:00","depth":13,"bounds":{"left":0.5415558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:30","depth":13,"bounds":{"left":0.55801195,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:00","depth":13,"bounds":{"left":0.57430184,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:30","depth":13,"bounds":{"left":0.59075797,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:00","depth":13,"bounds":{"left":0.6072141,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:30","depth":13,"bounds":{"left":0.6236702,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:00","depth":13,"bounds":{"left":0.64012635,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:30","depth":13,"bounds":{"left":0.6565825,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:00","depth":13,"bounds":{"left":0.67303854,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:30","depth":13,"bounds":{"left":0.68949467,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:00","depth":13,"bounds":{"left":0.70611703,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:30","depth":13,"bounds":{"left":0.72257316,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:00","depth":13,"bounds":{"left":0.73886305,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:30","depth":13,"bounds":{"left":0.7553192,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:00","depth":13,"bounds":{"left":0.77177525,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:30","depth":13,"bounds":{"left":0.7882314,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:00","depth":13,"bounds":{"left":0.804355,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:30","depth":13,"bounds":{"left":0.82081115,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:00","depth":13,"bounds":{"left":0.83759975,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"21:30","depth":13,"bounds":{"left":0.8540558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12 May 10:12 · Firefox / planhat - Google Search — Work","depth":10,"bounds":{"left":0.45628324,"top":0.2661612,"width":0.10488697,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"⏮ 30s","depth":9,"bounds":{"left":0.45694813,"top":0.7186752,"width":0.023936171,"height":0.02434158},"on_screen":true,"help_text":"Ctrl+←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"◀ 10s","depth":9,"bounds":{"left":0.48354387,"top":0.71907425,"width":0.02244016,"height":0.023942538},"on_screen":true,"help_text":"←","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"⏸ Pause","depth":9,"bounds":{"left":0.5086436,"top":0.7186752,"width":0.027925532,"height":0.02434158},"on_screen":true,"help_text":"Space","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"10s ▶","depth":9,"bounds":{"left":0.53922874,"top":0.71907425,"width":0.022273935,"height":0.023942538},"on_screen":true,"help_text":"→","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-335103884459386528
|
9209073691035053021
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30
21:00
21:30
12 May 10:12 · Firefox / planhat - Google Search — Work
⏮ 30s
◀ 10s
⏸ Pause
10s ▶...
|
32033
|
NULL
|
NULL
|
NULL
|
|
32036
|
1241
|
46
|
2026-05-13T09:03:57.209178+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663037209_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-6181084047028298607
|
8631489587908856471
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32037
|
1240
|
33
|
2026-05-13T09:03:58.433799+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663038433_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-6773229448344728745
|
8632612251553655765
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response...
|
32034
|
NULL
|
NULL
|
NULL
|
|
32038
|
1241
|
47
|
2026-05-13T09:03:58.401079+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663038401_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false}]...
|
8668322401747497175
|
8632612938731629533
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar...
|
32036
|
NULL
|
NULL
|
NULL
|
|
32039
|
1240
|
34
|
2026-05-13T09:03:59.672749+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663039672_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
151946306288913519
|
8488500362012666837
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32040
|
1241
|
48
|
2026-05-13T09:03:59.697368+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663039697_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32041
|
1240
|
35
|
2026-05-13T09:04:00.493825+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663040493_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit)....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-3597107170180916192
|
8491948436924034005
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit)....
|
32039
|
NULL
|
NULL
|
NULL
|
|
32042
|
1241
|
49
|
2026-05-13T09:04:00.490025+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663040490_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-633493124836282536
|
8487586665707760597
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said...
|
32040
|
NULL
|
NULL
|
NULL
|
|
32043
|
1240
|
36
|
2026-05-13T09:04:01.885384+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663041885_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
8618031042460022825
|
8488571830269520853
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown....
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32044
|
1241
|
50
|
2026-05-13T09:04:01.885396+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663041885_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32045
|
1240
|
37
|
2026-05-13T09:04:02.494855+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663042494_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
6840590589780677221
|
8488501461525343189
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language...
|
32043
|
NULL
|
NULL
|
NULL
|
|
32046
|
1240
|
38
|
2026-05-13T09:04:04.626045+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663044626_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-4500493459038834123
|
8487586657116761045
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32047
|
1241
|
51
|
2026-05-13T09:04:04.620988+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663044620_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-4408640441979658528
|
9209072993111256029
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people....
|
32044
|
NULL
|
NULL
|
NULL
|
|
32048
|
1240
|
39
|
2026-05-13T09:04:12.647867+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663052647_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
5758433181543905301
|
8632682609560398813
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in...
|
32046
|
NULL
|
NULL
|
NULL
|
|
32049
|
1241
|
52
|
2026-05-13T09:04:12.658971+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663052658_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32050
|
1240
|
40
|
2026-05-13T09:04:13.478899+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663053478_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"}]...
|
5338921852483979430
|
8487585695045151701
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32051
|
1240
|
41
|
2026-05-13T09:04:17.119693+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663057119_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4259861013841801623
|
8487585703635086293
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?...
|
32050
|
NULL
|
NULL
|
NULL
|
|
32052
|
1241
|
53
|
2026-05-13T09:04:17.102475+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663057102_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
6133219881644258259
|
8632612251528473565
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window...
|
32049
|
NULL
|
NULL
|
NULL
|
|
32053
|
1241
|
54
|
2026-05-13T09:04:18.502430+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663058502_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32054
|
1241
|
55
|
2026-05-13T09:04:24.570649+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663064570_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:30","depth":13,"bounds":{"left":0.45894283,"top":0.21947326,"width":0.00880984,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":13,"bounds":{"left":0.47573137,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:30","depth":13,"bounds":{"left":0.4921875,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:00","depth":13,"bounds":{"left":0.50880986,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:30","depth":13,"bounds":{"left":0.52526593,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:00","depth":13,"bounds":{"left":0.5415558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:30","depth":13,"bounds":{"left":0.55801195,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:00","depth":13,"bounds":{"left":0.57430184,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:30","depth":13,"bounds":{"left":0.59075797,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:00","depth":13,"bounds":{"left":0.6072141,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:30","depth":13,"bounds":{"left":0.6236702,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:00","depth":13,"bounds":{"left":0.64012635,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:30","depth":13,"bounds":{"left":0.6565825,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:00","depth":13,"bounds":{"left":0.67303854,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:30","depth":13,"bounds":{"left":0.68949467,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:00","depth":13,"bounds":{"left":0.70611703,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:30","depth":13,"bounds":{"left":0.72257316,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
428773019499955289
|
9209073141279239133
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30...
|
32053
|
NULL
|
NULL
|
NULL
|
|
32055
|
1241
|
56
|
2026-05-13T09:04:27.606130+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663067606_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:30","depth":13,"bounds":{"left":0.45894283,"top":0.21947326,"width":0.00880984,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":13,"bounds":{"left":0.47573137,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:30","depth":13,"bounds":{"left":0.4921875,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:00","depth":13,"bounds":{"left":0.50880986,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"11:30","depth":13,"bounds":{"left":0.52526593,"top":0.21947326,"width":0.0078125,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:00","depth":13,"bounds":{"left":0.5415558,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12:30","depth":13,"bounds":{"left":0.55801195,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:00","depth":13,"bounds":{"left":0.57430184,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"13:30","depth":13,"bounds":{"left":0.59075797,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:00","depth":13,"bounds":{"left":0.6072141,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"14:30","depth":13,"bounds":{"left":0.6236702,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:00","depth":13,"bounds":{"left":0.64012635,"top":0.21947326,"width":0.00831117,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15:30","depth":13,"bounds":{"left":0.6565825,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:00","depth":13,"bounds":{"left":0.67303854,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"16:30","depth":13,"bounds":{"left":0.68949467,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:00","depth":13,"bounds":{"left":0.70611703,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17:30","depth":13,"bounds":{"left":0.72257316,"top":0.21947326,"width":0.007978723,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:00","depth":13,"bounds":{"left":0.73886305,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"18:30","depth":13,"bounds":{"left":0.7553192,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:00","depth":13,"bounds":{"left":0.77177525,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"19:30","depth":13,"bounds":{"left":0.7882314,"top":0.21947326,"width":0.008144947,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:00","depth":13,"bounds":{"left":0.804355,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20:30","depth":13,"bounds":{"left":0.82081115,"top":0.21947326,"width":0.008643617,"height":0.008778931},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-3134105690210933431
|
9209073141279239133
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow
09:30
10:00
10:30
11:00
11:30
12:00
12:30
13:00
13:30
14:00
14:30
15:00
15:30
16:00
16:30
17:00
17:30
18:00
18:30
19:00
19:30
20:00
20:30...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32056
|
1241
|
57
|
2026-05-13T09:04:30.638428+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663070638_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Monitor","depth":9,"bounds":{"left":0.45262632,"top":0.10853951,"width":0.013464096,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jump to","depth":9,"bounds":{"left":0.8111702,"top":0.10853951,"width":0.01412899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.8312833,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":":","depth":9,"bounds":{"left":0.83710104,"top":0.10814046,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"--","depth":10,"bounds":{"left":0.84042555,"top":0.10814046,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Go","depth":8,"bounds":{"left":0.85920876,"top":0.10454908,"width":0.012300532,"height":0.018754989},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN","depth":10,"bounds":{"left":0.45761302,"top":0.14964086,"width":0.10571808,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"−","depth":9,"bounds":{"left":0.813996,"top":0.1452514,"width":0.009807181,"height":0.018754989},"on_screen":true,"help_text":"Zoom out","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"1×","depth":10,"bounds":{"left":0.82795876,"top":0.14924182,"width":0.004155585,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"+","depth":9,"bounds":{"left":0.83643615,"top":0.1452514,"width":0.009640957,"height":0.018754989},"on_screen":true,"help_text":"Zoom in","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Follow","depth":10,"bounds":{"left":0.8494016,"top":0.14924182,"width":0.004654255,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"checkbox","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Follow","depth":10,"bounds":{"left":0.85538566,"top":0.14924182,"width":0.011136968,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4869216098095446810
|
8632612388975815645
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
Monitor
Jump to
--
:
--
Go
APP TIMELINE · CLICK TO PLAY · DRAG SCROLLBAR TO PAN
−
1×
+
Follow
Follow...
|
32055
|
NULL
|
NULL
|
NULL
|
|
32057
|
1240
|
42
|
2026-05-13T09:04:31.617523+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663071617_m1.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"bounds":{"left":0.0,"top":0.21333334,"width":0.11666667,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.0,"top":0.29833335,"width":0.003125,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.06736111,"top":1.0,"width":0.027777778,"height":-0.053888917},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-9003491931173629337
|
8488712557020474325
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings....
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32058
|
1241
|
58
|
2026-05-13T09:04:31.604653+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663071604_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"AUDIO TRANSCRIPTIONS","depth":9,"bounds":{"left":0.45761302,"top":0.11532322,"width":0.04637633,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:28","depth":10,"bounds":{"left":0.4582779,"top":0.14086193,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.14405426,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's it.","depth":10,"bounds":{"left":0.48537233,"top":0.14126097,"width":0.018783245,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.16959298,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.17278531,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And now...","depth":10,"bounds":{"left":0.48520613,"top":0.16999201,"width":0.02244016,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.19832402,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.20151636,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-9089111667168672852
|
8632612938731629533
|
click
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32059
|
1241
|
59
|
2026-05-13T09:04:33.664845+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663073664_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"AUDIO TRANSCRIPTIONS","depth":9,"bounds":{"left":0.45761302,"top":0.11532322,"width":0.04637633,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:28","depth":10,"bounds":{"left":0.4582779,"top":0.14086193,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.14405426,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's it.","depth":10,"bounds":{"left":0.48537233,"top":0.14126097,"width":0.018783245,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.16959298,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.17278531,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And now...","depth":10,"bounds":{"left":0.48520613,"top":0.16999201,"width":0.02244016,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.19832402,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.20151636,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"bounds":{"left":0.48520613,"top":0.19872306,"width":0.023603724,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"bounds":{"left":0.4582779,"top":0.22705507,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.23024741,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Beijing","depth":10,"bounds":{"left":0.48537233,"top":0.22745411,"width":0.014295213,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"bounds":{"left":0.4582779,"top":0.25578612,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.25897846,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"causing concentreras.","depth":10,"bounds":{"left":0.48537233,"top":0.25618514,"width":0.047539894,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:35","depth":10,"bounds":{"left":0.4582779,"top":0.28451717,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.2877095,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"you","depth":10,"bounds":{"left":0.48520613,"top":0.2849162,"width":0.0078125,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:40","depth":10,"bounds":{"left":0.4582779,"top":0.31324822,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.31644055,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.","depth":10,"bounds":{"left":0.48537233,"top":0.31364724,"width":0.1783577,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:43","depth":10,"bounds":{"left":0.4582779,"top":0.34197924,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.3451716,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not gonna say that. I'm gonna say that.","depth":10,"bounds":{"left":0.48537233,"top":0.3423783,"width":0.09142287,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:44","depth":10,"bounds":{"left":0.4582779,"top":0.37071028,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.37390262,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"bounds":{"left":0.48537233,"top":0.37110934,"width":0.023769947,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:45","depth":10,"bounds":{"left":0.4582779,"top":0.39944133,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.40263367,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Let's hear. Do you want to run? Run.","depth":10,"bounds":{"left":0.48553857,"top":0.39984038,"width":0.07762633,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"bounds":{"left":0.4582779,"top":0.42817238,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.43136472,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.","depth":10,"bounds":{"left":0.48553857,"top":0.42857143,"width":0.19763963,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"bounds":{"left":0.4582779,"top":0.45690343,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.46009576,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.","depth":10,"bounds":{"left":0.48553857,"top":0.45730248,"width":0.3799867,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"bounds":{"left":0.4582779,"top":0.50239426,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.50558656,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.","depth":10,"bounds":{"left":0.48537233,"top":0.5027933,"width":0.3793218,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"bounds":{"left":0.4582779,"top":0.54788506,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.5510774,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.","depth":10,"bounds":{"left":0.48537233,"top":0.5482841,"width":0.37799203,"height":0.04708699},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"bounds":{"left":0.4582779,"top":0.6101357,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.61332804,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.","depth":10,"bounds":{"left":0.48553857,"top":0.6105347,"width":0.3465758,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"bounds":{"left":0.4582779,"top":0.6388667,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.6420591,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.","depth":10,"bounds":{"left":0.48553857,"top":0.6392658,"width":0.36353058,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"bounds":{"left":0.4582779,"top":0.6675978,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.6707901,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.","depth":10,"bounds":{"left":0.48553857,"top":0.6679968,"width":0.36901596,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"bounds":{"left":0.4582779,"top":0.7130886,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.71628094,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.","depth":10,"bounds":{"left":0.48553857,"top":0.7134876,"width":0.3715093,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"bounds":{"left":0.4582779,"top":0.7418196,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.745012,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.","depth":10,"bounds":{"left":0.48553857,"top":0.7422187,"width":0.37965426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"bounds":{"left":0.4582779,"top":0.7873105,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.7905028,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.","depth":10,"bounds":{"left":0.48553857,"top":0.7877095,"width":0.3799867,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.8328013,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.8359936,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.","depth":10,"bounds":{"left":0.48487368,"top":0.83320034,"width":0.2940492,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.86153233,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.86472464,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.","depth":10,"bounds":{"left":0.48487368,"top":0.8619314,"width":0.37400267,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.90702313,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.9102155,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.","depth":10,"bounds":{"left":0.48487368,"top":0.9074222,"width":0.37466756,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"bounds":{"left":0.4582779,"top":0.952514,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.9557063,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.","depth":10,"bounds":{"left":0.48537233,"top":0.952913,"width":0.38031915,"height":0.04708699},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010305851,"height":-0.014764547},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":1.0,"width":0.0068151597,"height":-0.017956853},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.","depth":10,"bounds":{"left":0.48537233,"top":1.0,"width":0.28956118,"height":-0.01516366},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010472074,"height":-0.043495655},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":1.0,"width":0.0068151597,"height":-0.04668796},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.","depth":10,"bounds":{"left":0.48553857,"top":1.0,"width":0.37749335,"height":-0.04389465},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010472074,"height":-0.0889864},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":1.0,"width":0.0068151597,"height":-0.09217882},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"same. I'm going to talk about the next episode.","depth":10,"bounds":{"left":0.48553857,"top":1.0,"width":0.10073138,"height":-0.08938551},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't have a lot of resources.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:58","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"because it's near.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yo, y'all, no connect, no connect.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What do you think about this?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not a British. Yes. I'm not sure what it is. I'm not a British.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Back for me because she was...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:07","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
1280558172880487358
|
-8364291012046904506
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one....
|
32058
|
NULL
|
NULL
|
NULL
|
|
32060
|
1241
|
60
|
2026-05-13T09:04:36.702074+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663076702_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense.
10:14
SYS
on the table.
10:20
SYS
Malitha.
10:20
SYS...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"AUDIO TRANSCRIPTIONS","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:28","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And now...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Beijing","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"causing concentreras.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:35","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"you","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:40","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:43","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not gonna say that. I'm gonna say that.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:44","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:45","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Let's hear. Do you want to run? Run.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"same. I'm going to talk about the next episode.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't have a lot of resources.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:58","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"because it's near.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yo, y'all, no connect, no connect.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What do you think about this?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not a British. Yes. I'm not sure what it is. I'm not a British.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Back for me because she was...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:07","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm almost there.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And where are the other two interplayers? I have a processing system.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:11","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use a little bit of water to prevent the water from sticking to the water.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry. Thank you very much. I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:14","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"didn't make some sense.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:14","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on the table.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:20","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Malitha.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:20","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-7961659890978486075
|
-6058447969010375866
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense.
10:14
SYS
on the table.
10:20
SYS
Malitha.
10:20
SYS...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32061
|
1241
|
61
|
2026-05-13T09:04:39.724033+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663079724_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061053474,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06264964,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.066640064,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.06464485,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06424581,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06424581,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06424581,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06424581,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06424581,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.06464485,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"AUDIO TRANSCRIPTIONS","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:28","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And now...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Beijing","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"causing concentreras.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:35","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"you","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:40","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:43","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not gonna say that. I'm gonna say that.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:44","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:45","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-1699805007315567094
|
9209073141279239133
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45...
|
32060
|
NULL
|
NULL
|
NULL
|
|
32062
|
1241
|
62
|
2026-05-13T09:04:42.785036+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663082785_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense.
10:14
SYS
on the table.
10:20
SYS
Malitha.
10:20
SYS
and such a pleasure.
10:21
SYS
For example, the payboard.
10:21
SYS
I mean, you met an event and it's an ed to buy a metronavent.
10:21
SYS
One cut, EI.
10:22
SYS
it does not lead to
10:22
SYS
The natives set up a lot.
10:23
SYS
good.
10:23
SYS
Thanks for watching.
10:23
SYS
and I'm going to see the other side of the robot
10:23
SYS
Thank you.
10:24...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.21747805,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.21907422,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.22306465,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.21628092,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.21628092,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.21628092,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.21628092,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.21628092,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.21628092,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.22106944,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.22067039,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.22067039,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.22067039,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.22067039,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.22067039,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.22106944,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"AUDIO TRANSCRIPTIONS","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:28","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And now...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Beijing","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"causing concentreras.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:35","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"you","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:40","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:43","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not gonna say that. I'm gonna say that.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:44","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:45","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Let's hear. Do you want to run? Run.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"same. I'm going to talk about the next episode.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't have a lot of resources.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:58","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"because it's near.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yo, y'all, no connect, no connect.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What do you think about this?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not a British. Yes. I'm not sure what it is. I'm not a British.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Back for me because she was...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:07","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm almost there.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And where are the other two interplayers? I have a processing system.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:11","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use a little bit of water to prevent the water from sticking to the water.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry. Thank you very much. I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:14","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"didn't make some sense.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:14","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on the table.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:20","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Malitha.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:20","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and such a pleasure.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:21","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"For example, the payboard.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:21","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, you met an event and it's an ed to buy a metronavent.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:21","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"One cut, EI.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:22","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"it does not lead to","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:22","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The natives set up a lot.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:23","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"good.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:23","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thanks for watching.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:23","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and I'm going to see the other side of the robot","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:23","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:24","depth":10,"bounds":{"left":0.4582779,"top":0.0,"width":0.009807181,"height":0.010774142},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
779418177036118993
|
-6058447969010375866
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense.
10:14
SYS
on the table.
10:20
SYS
Malitha.
10:20
SYS
and such a pleasure.
10:21
SYS
For example, the payboard.
10:21
SYS
I mean, you met an event and it's an ed to buy a metronavent.
10:21
SYS
One cut, EI.
10:22
SYS
it does not lead to
10:22
SYS
The natives set up a lot.
10:23
SYS
good.
10:23
SYS
Thanks for watching.
10:23
SYS
and I'm going to see the other side of the robot
10:23
SYS
Thank you.
10:24...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32063
|
1241
|
63
|
2026-05-13T09:04:45.831674+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663085831_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense.
10:14
SYS
on the table.
10:20
SYS
Malitha.
10:20
SYS
and such a pleasure....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.5422985,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.54389465,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.54788506,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.54070234,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.54070234,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.54070234,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.54070234,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.54070234,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.54070234,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.54588985,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.5454908,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.5454908,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.5454908,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.5454908,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.5454908,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.54588985,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"AUDIO TRANSCRIPTIONS","depth":9,"bounds":{"left":0.45761302,"top":0.11532322,"width":0.04637633,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:28","depth":10,"bounds":{"left":0.4582779,"top":0.14086193,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.14405426,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's it.","depth":10,"bounds":{"left":0.48537233,"top":0.14126097,"width":0.018783245,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.16959298,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.17278531,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And now...","depth":10,"bounds":{"left":0.48520613,"top":0.16999201,"width":0.02244016,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.19832402,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.20151636,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"bounds":{"left":0.48520613,"top":0.19872306,"width":0.023603724,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"bounds":{"left":0.4582779,"top":0.22705507,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.23024741,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Beijing","depth":10,"bounds":{"left":0.48537233,"top":0.22745411,"width":0.014295213,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"bounds":{"left":0.4582779,"top":0.25578612,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.25897846,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"causing concentreras.","depth":10,"bounds":{"left":0.48537233,"top":0.25618514,"width":0.047539894,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:35","depth":10,"bounds":{"left":0.4582779,"top":0.28451717,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.2877095,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"you","depth":10,"bounds":{"left":0.48520613,"top":0.2849162,"width":0.0078125,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:40","depth":10,"bounds":{"left":0.4582779,"top":0.31324822,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.31644055,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.","depth":10,"bounds":{"left":0.48537233,"top":0.31364724,"width":0.1783577,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:43","depth":10,"bounds":{"left":0.4582779,"top":0.34197924,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.3451716,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not gonna say that. I'm gonna say that.","depth":10,"bounds":{"left":0.48537233,"top":0.3423783,"width":0.09142287,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:44","depth":10,"bounds":{"left":0.4582779,"top":0.37071028,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.37390262,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"bounds":{"left":0.48537233,"top":0.37110934,"width":0.023769947,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:45","depth":10,"bounds":{"left":0.4582779,"top":0.39944133,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.40263367,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Let's hear. Do you want to run? Run.","depth":10,"bounds":{"left":0.48553857,"top":0.39984038,"width":0.07762633,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"bounds":{"left":0.4582779,"top":0.42817238,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.43136472,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.","depth":10,"bounds":{"left":0.48553857,"top":0.42857143,"width":0.19763963,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"bounds":{"left":0.4582779,"top":0.45690343,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.46009576,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.","depth":10,"bounds":{"left":0.48553857,"top":0.45730248,"width":0.3799867,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"bounds":{"left":0.4582779,"top":0.50239426,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.50558656,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.","depth":10,"bounds":{"left":0.48537233,"top":0.5027933,"width":0.3793218,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"bounds":{"left":0.4582779,"top":0.54788506,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.5510774,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.","depth":10,"bounds":{"left":0.48537233,"top":0.5482841,"width":0.37799203,"height":0.04708699},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"bounds":{"left":0.4582779,"top":0.6101357,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.61332804,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.","depth":10,"bounds":{"left":0.48553857,"top":0.6105347,"width":0.3465758,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"bounds":{"left":0.4582779,"top":0.6388667,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.6420591,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.","depth":10,"bounds":{"left":0.48553857,"top":0.6392658,"width":0.36353058,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"bounds":{"left":0.4582779,"top":0.6675978,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.6707901,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.","depth":10,"bounds":{"left":0.48553857,"top":0.6679968,"width":0.36901596,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"bounds":{"left":0.4582779,"top":0.7130886,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.71628094,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.","depth":10,"bounds":{"left":0.48553857,"top":0.7134876,"width":0.3715093,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"bounds":{"left":0.4582779,"top":0.7418196,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.745012,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.","depth":10,"bounds":{"left":0.48553857,"top":0.7422187,"width":0.37965426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"bounds":{"left":0.4582779,"top":0.7873105,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.7905028,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.","depth":10,"bounds":{"left":0.48553857,"top":0.7877095,"width":0.3799867,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.8328013,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.8359936,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.","depth":10,"bounds":{"left":0.48487368,"top":0.83320034,"width":0.2940492,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.86153233,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.86472464,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.","depth":10,"bounds":{"left":0.48487368,"top":0.8619314,"width":0.37400267,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.90702313,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.9102155,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.","depth":10,"bounds":{"left":0.48487368,"top":0.9074222,"width":0.37466756,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"bounds":{"left":0.4582779,"top":0.952514,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.9557063,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.","depth":10,"bounds":{"left":0.48537233,"top":0.952913,"width":0.38031915,"height":0.04708699},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010305851,"height":-0.014764547},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":1.0,"width":0.0068151597,"height":-0.017956853},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.","depth":10,"bounds":{"left":0.48537233,"top":1.0,"width":0.28956118,"height":-0.01516366},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010472074,"height":-0.043495655},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":1.0,"width":0.0068151597,"height":-0.04668796},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.","depth":10,"bounds":{"left":0.48553857,"top":1.0,"width":0.37749335,"height":-0.04389465},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010472074,"height":-0.0889864},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":1.0,"width":0.0068151597,"height":-0.09217882},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"same. I'm going to talk about the next episode.","depth":10,"bounds":{"left":0.48553857,"top":1.0,"width":0.10073138,"height":-0.08938551},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't have a lot of resources.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:58","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"because it's near.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yo, y'all, no connect, no connect.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What do you think about this?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not a British. Yes. I'm not sure what it is. I'm not a British.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Back for me because she was...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:07","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm almost there.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And where are the other two interplayers? I have a processing system.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:11","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use a little bit of water to prevent the water from sticking to the water.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry. Thank you very much. I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:14","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"didn't make some sense.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:14","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on the table.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:20","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Malitha.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:20","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and such a pleasure.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-3851100741238532448
|
-6058447969010375866
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense.
10:14
SYS
on the table.
10:20
SYS
Malitha.
10:20
SYS
and such a pleasure....
|
32062
|
NULL
|
NULL
|
NULL
|
|
32064
|
1241
|
64
|
2026-05-13T09:04:48.844733+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663088844_m2.jpg...
|
Firefox
|
Screenpipe — Archive — Personal
|
1
|
app.screenpipe.lakylak.xyz
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.0518755,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.06304868,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"All docs · AFFiNE","depth":4,"bounds":{"left":0.0,"top":0.08459697,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All docs · AFFiNE","depth":5,"bounds":{"left":0.013297873,"top":0.09577015,"width":0.029587766,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"DXP4800PLUS-B5F8","depth":4,"bounds":{"left":0.0,"top":0.11731844,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"DXP4800PLUS-B5F8","depth":5,"bounds":{"left":0.013297873,"top":0.12849163,"width":0.036901597,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.0,"top":0.15003991,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.013297873,"top":0.16121309,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Screenpipe — Archive","depth":4,"bounds":{"left":0.0,"top":0.18276137,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Screenpipe — Archive","depth":5,"bounds":{"left":0.013297873,"top":0.19393456,"width":0.037898935,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.05651596,"top":0.18994413,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"SQLite Web: archive.db","depth":4,"bounds":{"left":0.0,"top":0.21548285,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: archive.db","depth":5,"bounds":{"left":0.013297873,"top":0.22665602,"width":0.040724736,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"SQLite Web: db.sqlite","depth":4,"bounds":{"left":0.0,"top":0.2482043,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"SQLite Web: db.sqlite","depth":5,"bounds":{"left":0.013297873,"top":0.25937748,"width":0.03756649,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Claude","depth":4,"bounds":{"left":0.0,"top":0.28092578,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Claude","depth":5,"bounds":{"left":0.013297873,"top":0.29209897,"width":0.012134309,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":4,"bounds":{"left":0.0,"top":0.31364724,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Manage extra usage for paid Claude plans | Claude Help Center","depth":5,"bounds":{"left":0.013297873,"top":0.32482043,"width":0.1100399,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"2 TB in 25 MB/s - Google Search","depth":4,"bounds":{"left":0.0,"top":0.3463687,"width":0.06881649,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 TB in 25 MB/s - Google Search","depth":5,"bounds":{"left":0.013297873,"top":0.3575419,"width":0.05668218,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.0028257978,"top":0.38068634,"width":0.06333112,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0028257978,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.013796543,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.024933511,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.036070477,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bitwarden","depth":6,"bounds":{"left":0.04720745,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.29321808,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.30518618,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google Account: Lukáš Koválik (kovaliklukas@gmail.com)","depth":12,"bounds":{"left":0.3025266,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.07280585,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.08610372,"top":0.10454908,"width":0.028590426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini","depth":15,"bounds":{"left":0.0887633,"top":0.10973663,"width":0.021941489,"height":0.020351157},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"New chat","depth":12,"bounds":{"left":0.2613032,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share conversation","depth":11,"bounds":{"left":0.27460107,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.28789893,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.068484046,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.068484046,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe is fundamentally designed to be a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"100% local, privacy-first","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":27,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"1. Where It Starts: The Capture Layer","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Where It Starts: The Capture Layer","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"System Audio:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Microphone Input:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Everything you say into your mic.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. How It Gets Transferred and Transcribed","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. How It Gets Transferred and Transcribed","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe prioritizes local processing, the \"transfer\" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is the transcription pipeline:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Engine:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The raw audio chunks are fed into a local speech-to-text model.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OpenAI Whisper","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Diarization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As it transcribes the text, the engine also performs \"diarization\"—a technical term for speaker identification.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"View source details. Opens side panel.","depth":29,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Storage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The final transcribed text is then indexed into a local","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SQLite database","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The \"Work in Progress\" (WIP) Stage","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The \"Work in Progress\" (WIP) Stage","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There is a slight bottleneck between","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"hearing","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the audio and","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"saving","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"the text, which acts as the WIP stage:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buffering:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Processing Queue:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Finalization:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"4. Understanding the Folder Structure","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"4. Understanding the Folder Structure","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The SQLite Database (","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.05817819,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"):","depth":29,"bounds":{"left":0.17885639,"top":0.0,"width":0.0034906915,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is \"done.\"","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21010639,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.011303191,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"bounds":{"left":0.120678194,"top":0.0,"width":0.008144947,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folders:","depth":29,"bounds":{"left":0.14960106,"top":0.0,"width":0.020944148,"height":0.016360734},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.","depth":29,"bounds":{"left":0.09142287,"top":0.0,"width":0.21542554,"height":0.037110932},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Temp Files:","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.028922873,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you see rapidly changing files, temporary","depth":29,"bounds":{"left":0.12034574,"top":0.029928172,"width":0.106715426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"chunks, or locked database journals (like","depth":29,"bounds":{"left":0.09142287,"top":0.029928172,"width":0.21991356,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.","depth":29,"bounds":{"left":0.09142287,"top":0.050678372,"width":0.22174202,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?","depth":27,"bounds":{"left":0.0787899,"top":0.13288109,"width":0.22456782,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sources","depth":26,"bounds":{"left":0.0787899,"top":0.18715084,"width":0.036402926,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sources","depth":28,"bounds":{"left":0.091755316,"top":0.19592977,"width":0.017785905,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"bounds":{"left":0.075465426,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"bounds":{"left":0.08610372,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"bounds":{"left":0.09674202,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"bounds":{"left":0.107380316,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"bounds":{"left":0.11801862,"top":0.23184358,"width":0.010638298,"height":0.025538707},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.14029256,"top":0.292498,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":21,"bounds":{"left":0.16023937,"top":0.30207503,"width":0.13696809,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.068484046,"top":0.3028731,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?","depth":23,"bounds":{"left":0.16023937,"top":0.30367118,"width":0.13663563,"height":0.13128492},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.29720744,"top":0.30207503,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":24,"bounds":{"left":0.3025266,"top":0.42976856,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"bounds":{"left":0.09208777,"top":0.43216282,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"bounds":{"left":0.08976064,"top":0.4736632,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"bounds":{"left":0.08976064,"top":0.47565842,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yes, those","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.025930852,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.106715426,"top":0.47805268,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.","depth":27,"bounds":{"left":0.0787899,"top":0.47685555,"width":0.23005319,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is exactly what those files are doing sitting in your","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.1341423,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"bounds":{"left":0.21492687,"top":0.528332,"width":0.053025264,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"directory and how they interact with the rest of the system.","depth":27,"bounds":{"left":0.0787899,"top":0.5271349,"width":0.22623006,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. The \"Time Machine\" Playback","depth":26,"bounds":{"left":0.0787899,"top":0.5885874,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. The \"Time Machine\" Playback","depth":27,"bounds":{"left":0.0787899,"top":0.59018356,"width":0.08194814,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the","depth":27,"bounds":{"left":0.0787899,"top":0.61652035,"width":0.20761304,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"bounds":{"left":0.24883644,"top":0.63846767,"width":0.025099734,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"database along with","depth":27,"bounds":{"left":0.0787899,"top":0.63727057,"width":0.22240691,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"precise timestamps and file path references","depth":27,"bounds":{"left":0.105053194,"top":0.65802073,"width":0.11419548,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"bounds":{"left":0.21924867,"top":0.65802073,"width":0.0013297872,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific","depth":27,"bounds":{"left":0.0787899,"top":0.6875499,"width":0.22755983,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.27293882,"top":0.7094972,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.","depth":27,"bounds":{"left":0.0787899,"top":0.70830005,"width":0.23354389,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. Re-processing and Model Upgrades","depth":26,"bounds":{"left":0.0787899,"top":0.811253,"width":0.234375,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. Re-processing and Model Upgrades","depth":27,"bounds":{"left":0.0787899,"top":0.81284916,"width":0.09840426,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw","depth":27,"bounds":{"left":0.0787899,"top":0.83918595,"width":0.23254654,"height":0.07861133},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.117519945,"top":0.90263367,"width":0.011136968,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.","depth":27,"bounds":{"left":0.0787899,"top":0.90143657,"width":0.23038563,"height":0.057861134},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. The Source of Truth for Hallucinations","depth":26,"bounds":{"left":0.0787899,"top":0.98363924,"width":0.234375,"height":0.01636076},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. The Source of Truth for Hallucinations","depth":27,"bounds":{"left":0.0787899,"top":0.98523545,"width":0.10405585,"height":0.014764547},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.2209109,"height":-0.011572242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".mp4","depth":28,"bounds":{"left":0.14660904,"top":1.0,"width":0.011303191,"height":-0.05426979},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"file is your fallback source of truth to verify what was actually said.","depth":27,"bounds":{"left":0.0787899,"top":1.0,"width":0.23121676,"height":-0.05307257},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Database Views Them","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Database Views Them","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you were to open up your","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"db.sqlite","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and query the tables (e.g.,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"audio_transcriptions","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), you would see that the database treats the","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/data/","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"soundcore AeroClip (input)_2026-05-12_07-40-48.mp4","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What happens if you delete them?","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you manually","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"rm","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a \"file not found\" error in the background logs.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Managing the Storage Footprint","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Managing the Storage Footprint","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Because ScreenPipe records continuously, this folder will inevitably grow over time.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Built-in Garbage Collection:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Archiving:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you prefer to keep a permanent, searchable \"life log\" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"older","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":24,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share and export","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":24,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said yes I will do that. Is there a way to setup languages to transcription?","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes I will do that. Is there a way to setup languages to transcription?","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":24,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"As for setting up languages for transcription,","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"yes, you can configure it","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", but it helps to understand how ScreenPipe handles language natively first.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How the Transcription Engine Handles Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How the Transcription Engine Handles Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By default, ScreenPipe uses a","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"multilingual Whisper model","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and sets the language configuration to","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-Detect","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Advantage:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The Drawback:","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How to Force a Specific Language","depth":26,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How to Force a Specific Language","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are using the Desktop App UI:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Open the ScreenPipe settings.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Navigate to the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Audio","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI/Transcription","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"section.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Look for the","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Whisper Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Transcription Language","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"dropdown.","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Change it from \"Auto\" to your specific language (e.g., English, Bulgarian, or Slovak).","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you are running ScreenPipe via CLI/Config:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You can modify your underlying configuration (usually found in","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"~/.screenpipe/pipe.json","depth":28,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Slovak)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(Bulgarian)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(English)","depth":29,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":23,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":23,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share and export","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":23,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Ask Gemini","depth":20,"bounds":{"left":0.08211436,"top":0.83439744,"width":0.22573139,"height":0.01915403},"on_screen":true,"value":"Ask Gemini","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Ask Gemini","depth":21,"bounds":{"left":0.08211436,"top":0.8347965,"width":0.030086435,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.078125,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.094082445,"top":0.87031126,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.27044547,"top":0.867917,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.2757646,"top":0.87669593,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.29853722,"top":0.867917,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.30485374,"top":0.8671189,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Gemini is AI and can make mistakes, including about people.","depth":17,"bounds":{"left":0.11702128,"top":0.92178774,"width":0.11170213,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy and Gemini Opens in a new window","depth":17,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy and Gemini","depth":18,"bounds":{"left":0.2287234,"top":0.92178774,"width":0.044215426,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.068484046,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.07413564,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.07978723,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Screenpipe [archive.db · 763MB]","depth":7,"bounds":{"left":0.33061835,"top":0.061452515,"width":0.06067154,"height":0.017956903},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Screenpipe","depth":8,"bounds":{"left":0.33061835,"top":0.06304868,"width":0.027759308,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"[archive.db · 763MB]","depth":9,"bounds":{"left":0.35970744,"top":0.06703911,"width":0.03158245,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Activity","depth":7,"bounds":{"left":0.39594415,"top":0.059856344,"width":0.024933511,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":7,"bounds":{"left":0.42154256,"top":0.059856344,"width":0.023603724,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Audio","depth":7,"bounds":{"left":0.44581118,"top":0.059856344,"width":0.020944148,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Work Report","depth":7,"bounds":{"left":0.46742022,"top":0.059856344,"width":0.03507314,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Timetable","depth":7,"bounds":{"left":0.5031583,"top":0.059856344,"width":0.029753989,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Summary","depth":7,"bounds":{"left":0.53357714,"top":0.059856344,"width":0.034075797,"height":0.0207502},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Date","depth":8,"bounds":{"left":0.93866354,"top":0.0650439,"width":0.008144947,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"12","depth":9,"bounds":{"left":0.95545214,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.96127,"top":0.06464485,"width":0.0023271276,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"05","depth":9,"bounds":{"left":0.9645944,"top":0.06464485,"width":0.0048204786,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"/","depth":8,"bounds":{"left":0.97041225,"top":0.06464485,"width":0.002493351,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2026","depth":9,"bounds":{"left":0.97390294,"top":0.06464485,"width":0.009474734,"height":0.011572227},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":8,"bounds":{"left":0.9847075,"top":0.0650439,"width":0.0051529254,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"AUDIO TRANSCRIPTIONS","depth":9,"bounds":{"left":0.45761302,"top":0.11532322,"width":0.04637633,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:28","depth":10,"bounds":{"left":0.4582779,"top":0.14086193,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.14405426,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's it.","depth":10,"bounds":{"left":0.48537233,"top":0.14126097,"width":0.018783245,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.16959298,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.17278531,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And now...","depth":10,"bounds":{"left":0.48520613,"top":0.16999201,"width":0.02244016,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:32","depth":10,"bounds":{"left":0.4582779,"top":0.19832402,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.20151636,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"bounds":{"left":0.48520613,"top":0.19872306,"width":0.023603724,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"bounds":{"left":0.4582779,"top":0.22705507,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.23024741,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Beijing","depth":10,"bounds":{"left":0.48537233,"top":0.22745411,"width":0.014295213,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:33","depth":10,"bounds":{"left":0.4582779,"top":0.25578612,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.25897846,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"causing concentreras.","depth":10,"bounds":{"left":0.48537233,"top":0.25618514,"width":0.047539894,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:35","depth":10,"bounds":{"left":0.4582779,"top":0.28451717,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.47357047,"top":0.2877095,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"you","depth":10,"bounds":{"left":0.48520613,"top":0.2849162,"width":0.0078125,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:40","depth":10,"bounds":{"left":0.4582779,"top":0.31324822,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.31644055,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.","depth":10,"bounds":{"left":0.48537233,"top":0.31364724,"width":0.1783577,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:43","depth":10,"bounds":{"left":0.4582779,"top":0.34197924,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.3451716,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not gonna say that. I'm gonna say that.","depth":10,"bounds":{"left":0.48537233,"top":0.3423783,"width":0.09142287,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:44","depth":10,"bounds":{"left":0.4582779,"top":0.37071028,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"bounds":{"left":0.4737367,"top":0.37390262,"width":0.0066489363,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Thank you.","depth":10,"bounds":{"left":0.48537233,"top":0.37110934,"width":0.023769947,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:45","depth":10,"bounds":{"left":0.4582779,"top":0.39944133,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.40263367,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Let's hear. Do you want to run? Run.","depth":10,"bounds":{"left":0.48553857,"top":0.39984038,"width":0.07762633,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"bounds":{"left":0.4582779,"top":0.42817238,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.43136472,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.","depth":10,"bounds":{"left":0.48553857,"top":0.42857143,"width":0.19763963,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:46","depth":10,"bounds":{"left":0.4582779,"top":0.45690343,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.46009576,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.","depth":10,"bounds":{"left":0.48553857,"top":0.45730248,"width":0.3799867,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"bounds":{"left":0.4582779,"top":0.50239426,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.50558656,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.","depth":10,"bounds":{"left":0.48537233,"top":0.5027933,"width":0.3793218,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:47","depth":10,"bounds":{"left":0.4582779,"top":0.54788506,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.5510774,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.","depth":10,"bounds":{"left":0.48537233,"top":0.5482841,"width":0.37799203,"height":0.04708699},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"bounds":{"left":0.4582779,"top":0.6101357,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.61332804,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.","depth":10,"bounds":{"left":0.48553857,"top":0.6105347,"width":0.3465758,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:48","depth":10,"bounds":{"left":0.4582779,"top":0.6388667,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.6420591,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.","depth":10,"bounds":{"left":0.48553857,"top":0.6392658,"width":0.36353058,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"bounds":{"left":0.4582779,"top":0.6675978,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.6707901,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.","depth":10,"bounds":{"left":0.48553857,"top":0.6679968,"width":0.36901596,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:49","depth":10,"bounds":{"left":0.4582779,"top":0.7130886,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.71628094,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.","depth":10,"bounds":{"left":0.48553857,"top":0.7134876,"width":0.3715093,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"bounds":{"left":0.4582779,"top":0.7418196,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.745012,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.","depth":10,"bounds":{"left":0.48553857,"top":0.7422187,"width":0.37965426,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:50","depth":10,"bounds":{"left":0.4582779,"top":0.7873105,"width":0.010472074,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":0.7905028,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.","depth":10,"bounds":{"left":0.48553857,"top":0.7877095,"width":0.3799867,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.8328013,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.8359936,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.","depth":10,"bounds":{"left":0.48487368,"top":0.83320034,"width":0.2940492,"height":0.013567438},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.86153233,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.86472464,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.","depth":10,"bounds":{"left":0.48487368,"top":0.8619314,"width":0.37400267,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:51","depth":10,"bounds":{"left":0.4582779,"top":0.90702313,"width":0.009807181,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4730718,"top":0.9102155,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.","depth":10,"bounds":{"left":0.48487368,"top":0.9074222,"width":0.37466756,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"bounds":{"left":0.4582779,"top":0.952514,"width":0.010305851,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":0.9557063,"width":0.0068151597,"height":0.009976057},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.","depth":10,"bounds":{"left":0.48537233,"top":0.952913,"width":0.38031915,"height":0.04708699},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:52","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010305851,"height":-0.014764547},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.47357047,"top":1.0,"width":0.0068151597,"height":-0.017956853},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.","depth":10,"bounds":{"left":0.48537233,"top":1.0,"width":0.28956118,"height":-0.01516366},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010472074,"height":-0.043495655},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":1.0,"width":0.0068151597,"height":-0.04668796},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.","depth":10,"bounds":{"left":0.48553857,"top":1.0,"width":0.37749335,"height":-0.04389465},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:53","depth":10,"bounds":{"left":0.4582779,"top":1.0,"width":0.010472074,"height":-0.0889864},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"bounds":{"left":0.4737367,"top":1.0,"width":0.0068151597,"height":-0.09217882},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"same. I'm going to talk about the next episode.","depth":10,"bounds":{"left":0.48553857,"top":1.0,"width":0.10073138,"height":-0.08938551},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:54","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't have a lot of resources.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:55","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not sure if you can see it. I'm not sure if you can see it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:56","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:57","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:58","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"09:59","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:00","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"because it's near.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Yo, y'all, no connect, no connect.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:01","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:02","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:03","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:04","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What do you think about this?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:05","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm not a British. Yes. I'm not sure what it is. I'm not a British.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Back for me because she was...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:06","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:07","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then?","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:08","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:09","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm almost there.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And where are the other two interplayers? I have a processing system.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:10","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I think it's a good idea. I think it's a good idea.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:11","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MIC","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm gonna use a little bit of water to prevent the water from sticking to the water.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:12","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I'm sorry. Thank you very much. I'm sorry.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:13","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"10:14","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SYS","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"didn't make some sense.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
6049806292030385319
|
-6058448003370114234
|
visual_change
|
accessibility
|
NULL
|
Screenpipe — Archive
Screenpipe — Archive
All docs Screenpipe — Archive
Screenpipe — Archive
All docs · AFFiNE
All docs · AFFiNE
DXP4800PLUS-B5F8
DXP4800PLUS-B5F8
New Tab
New Tab
Screenpipe — Archive
Screenpipe — Archive
Close tab
SQLite Web: archive.db
SQLite Web: archive.db
SQLite Web: db.sqlite
SQLite Web: db.sqlite
Claude
Claude
Manage extra usage for paid Claude plans | Claude Help Center
Manage extra usage for paid Claude plans | Claude Help Center
2 TB in 25 MB/s - Google Search
2 TB in 25 MB/s - Google Search
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Open history (⇧⌘H)
Open bookmarks (⌘B)
Bitwarden
AI Chat settings
Close
Google Account: Lukáš Koválik ([EMAIL])
Main menu
New chat
Gemini
New chat
Share conversation
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
You said
Exaplin me: how does the ScreenPipe audio recording work? Where did it start? How is it getting transferred and transcribed? Is there some work in progress stage? How can I see from the folder structure what has been done and what hasn't been done?
Listen
Show thinking
Gemini said
Gemini said
Here is a breakdown of how ScreenPipe handles audio recording, from the moment it captures sound to how it organizes the data on your machine.
ScreenPipe is fundamentally designed to be a
100% local, privacy-first
application, meaning the vast majority of its processing happens directly on your hardware without relying on the cloud.
View source details. Opens side panel.
1. Where It Starts: The Capture Layer
1. Where It Starts: The Capture Layer
The process begins locally by tapping directly into your operating system’s audio routing. ScreenPipe simultaneously captures two separate audio streams:
System Audio:
Everything your computer outputs (e.g., the voices of other people in a Zoom or Teams meeting, a YouTube video, or Spotify).
Microphone Input:
Everything you say into your mic.
View source details. Opens side panel.
Because ScreenPipe acts as a 24/7 background process, it continuously listens to these streams and breaks the continuous audio into manageable chunks to prepare them for the next stage.
2. How It Gets Transferred and Transcribed
2. How It Gets Transferred and Transcribed
Because ScreenPipe prioritizes local processing, the "transfer" step is incredibly short. The audio data never leaves your machine unless you explicitly set up a cloud plugin.
Here is the transcription pipeline:
The Engine:
The raw audio chunks are fed into a local speech-to-text model.
By default, ScreenPipe uses
OpenAI Whisper
running locally on your hardware. (Users can also configure it to use cloud providers like Deepgram if they need faster processing, but local Whisper is the standard).
View source details. Opens side panel.
Diarization:
As it transcribes the text, the engine also performs "diarization"—a technical term for speaker identification.
It analyzes the audio to distinguish between your voice and the voices of others, labeling who said what.
View source details. Opens side panel.
Storage:
The final transcribed text is then indexed into a local
SQLite database
. This database uses Full-Text Search (FTS5), which is what allows you to instantly search for a phrase you heard in a meeting three weeks ago.
3. The "Work in Progress" (WIP) Stage
3. The "Work in Progress" (WIP) Stage
There is a slight bottleneck between
hearing
the audio and
saving
the text, which acts as the WIP stage:
Buffering:
Audio is recorded into a temporary buffer in your system's RAM or written as temporary chunk files on your disk.
Processing Queue:
Whisper requires CPU/GPU resources to transcribe. If you are having a rapid, multi-person conversation, the raw audio chunks line up in a processing queue.
Finalization:
Once the AI finishes transcribing a chunk, the text is committed to the SQLite database alongside a timestamp, and the raw audio is compressed and saved so you can play it back later.
4. Understanding the Folder Structure
4. Understanding the Folder Structure
All of ScreenPipe's data lives locally on your machine (usually in a hidden directory like
~/.screenpipe
on Mac/Linux or in your AppData folder on Windows). You can determine what has been processed by looking at a few key areas in this directory:
The SQLite Database (
):
This is the master ledger. If text exists inside this database, it means the audio has been fully transcribed, diarized, and is "done."
The
or
folders:
This is where the compressed, raw audio files and JPEG screenshots are stored permanently. Think of this as the raw archive.
Temp Files:
If you see rapidly changing files, temporary
chunks, or locked database journals (like
), this indicates the system is actively processing a backlog of audio (the WIP stage). Once the engine catches up, these temporary chunk files are usually cleared out or moved to permanent storage, and the database updates.
Are you setting up ScreenPipe for a specific workflow (like summarizing meetings or tracking your daily tasks), or just exploring how the architecture works under the hood?
Sources
Sources
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
You said I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
You said
I see the files in ~/.screenpipe/data/ as LakyLak bose qc35 II (input)_2026-05-12_06-49-17.mp4 or MacBook Pro Microphone (input)_2026-05-12_12-17-23.mp4 or soundcore AeroClip (input)_2026-05-12_07-40-48.mp4 or System Audio (output)_2026-05-11_06-17-14.mp4. Are these used for anything after transcribing and storing in sqlite db?
Expand
Listen
Show thinking
Gemini said
Gemini said
Yes, those
.mp4
files are absolutely still used after the transcription process is complete. They are not temporary processing chunks; they act as your permanent media archive.
Here is exactly what those files are doing sitting in your
~/.screenpipe/data/
directory and how they interact with the rest of the system.
1. The "Time Machine" Playback
1. The "Time Machine" Playback
The primary reason ScreenPipe keeps these files is for audio playback. When Whisper transcribes your meetings or ambient audio, it writes the text into the
db.sqlite
database along with
precise timestamps and file path references
.
When you search your ScreenPipe history for a specific conversation, the UI doesn't just show you the text—it uses the file path stored in the database to pull up that specific
.mp4
file and starts playing it at the exact second the keyword was spoken. Without these files, you get the transcript, but you lose the ability to actually listen to the context, tone, or nuance of the original conversation.
2. Re-processing and Model Upgrades
2. Re-processing and Model Upgrades
Local LLMs and transcription models are improving rapidly. Right now, your system might be using a quantized Whisper model to balance CPU load with accuracy. If you later decide to run a heavier, more accurate model—or if a radically better speaker diarization algorithm is released—having the raw
.mp4
files allows you to re-process historical audio. Without the raw input files, your data is permanently locked to the quality of the transcription model you were running on May 12th.
3. The Source of Truth for Hallucinations
3. The Source of Truth for Hallucinations
Speech-to-text engines, even great ones, occasionally hallucinate or misinterpret technical jargon, overlapping voices, or background noise. If a database query returns a completely nonsensical sentence, that
.mp4
file is your fallback source of truth to verify what was actually said.
How the Database Views Them
How the Database Views Them
If you were to open up your
db.sqlite
and query the tables (e.g.,
audio_transcriptions
), you would see that the database treats the
~/.screenpipe/data/
folder essentially as a blob storage backend. The tables contain columns acting as soft foreign keys pointing to filenames like
soundcore AeroClip (input)_2026-05-12_07-40-48.mp4
.
What happens if you delete them?
If you manually
rm
those files to free up space on your Mac's SSD, ScreenPipe won't crash. The SQLite database will still hold all your searchable text. However, any attempt to play back the audio for those older transcripts will silently fail or throw a "file not found" error in the background logs.
Managing the Storage Footprint
Managing the Storage Footprint
Because ScreenPipe records continuously, this folder will inevitably grow over time.
Built-in Garbage Collection:
ScreenPipe usually handles this automatically based on retention settings (e.g., deleting media older than 30 days or when the folder hits a specific GB limit).
Archiving:
If you prefer to keep a permanent, searchable "life log" without eating up your local SSD space, you don't necessarily have to let ScreenPipe delete them. You could script a simple cron job to regularly
older
files to your massive 4-drive RAID 5 array, replacing the local files with symlinks. This would keep the database references intact while offloading the heavy media storage to your NAS.
Good response
Bad response
Share and export
Copy
Show more options
Copy prompt
Edit
You said yes I will do that. Is there a way to setup languages to transcription?
You said
yes I will do that. Is there a way to setup languages to transcription?
Listen
Show thinking
Gemini said
Gemini said
That is a solid plan. Offloading those older media files to your 4-drive RAID 5 setup (or even the secondary SSD-only Beelink NAS if you end up building it) will keep your local SSD lean while keeping your database searchable.
As for setting up languages for transcription,
yes, you can configure it
, but it helps to understand how ScreenPipe handles language natively first.
How the Transcription Engine Handles Language
How the Transcription Engine Handles Language
By default, ScreenPipe uses a
multilingual Whisper model
and sets the language configuration to
Auto-Detect
.
When a new audio chunk comes in, the model listens to the first few seconds of audio, guesses the language, and transcribes the rest of that chunk using that assumption.
The Advantage:
If you are watching a tutorial in English, and then an hour later you are in a Zoom call speaking Bulgarian, ScreenPipe will automatically switch gears without you having to touch the settings.
The Drawback:
Auto-detect can sometimes guess wrong, especially if there is heavy background noise, if someone starts a sentence with a foreign loanword, or if you are rapidly code-switching between Slovak and English in the same conversation. When it guesses wrong, it might try to force Slovak words into English spelling, resulting in gibberish.
How to Force a Specific Language
How to Force a Specific Language
If you want to ensure maximum accuracy and you know you will primarily be recording in one language for a specific period, you can explicitly set the language. This also slightly speeds up the processing because the model skips the detection phase.
If you are using the Desktop App UI:
Open the ScreenPipe settings.
Navigate to the
Audio
or
AI/Transcription
section.
Look for the
Whisper Language
or
Transcription Language
dropdown.
Change it from "Auto" to your specific language (e.g., English, Bulgarian, or Slovak).
If you are running ScreenPipe via CLI/Config:
You can modify your underlying configuration (usually found in
~/.screenpipe/pipe.json
or by passing a flag when starting the daemon) to force a specific language code. You would use the standard ISO language codes:
(Slovak)
(Bulgarian)
(English)
If you find yourself frequently jumping between those three languages, your best bet is to leave it on Auto-Detect and perhaps run a slightly larger, more accurate Whisper model (like Whisper 'Small' or 'Base' instead of 'Tiny') if your hardware has the headroom for it.
Good response
Bad response
Redo
Share and export
Copy
Show more options
Ask Gemini
Ask Gemini
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Gemini is AI and can make mistakes, including about people.
Your privacy and Gemini Opens in a new window
Your privacy and Gemini
Opens in a new window
Summarize page
Summarize page
Screenpipe [archive.db · 763MB]
Screenpipe
[archive.db · 763MB]
Activity
Search
Audio
Work Report
Timetable
AI Summary
Date
12
/
05
/
2026
Calendar
AUDIO TRANSCRIPTIONS
09:28
MIC
That's it.
09:32
MIC
And now...
09:32
MIC
Thank you.
09:33
MIC
Beijing
09:33
MIC
causing concentreras.
09:35
MIC
you
09:40
MIC
I'm gonna use the handkerchief to make sure that the handkerchief is not too thick.
09:43
MIC
I'm not gonna say that. I'm gonna say that.
09:44
MIC
Thank you.
09:45
SYS
Let's hear. Do you want to run? Run.
09:46
SYS
I'm going to show you. I'm going to show you. I'm going to show you. I'm going to show you.
09:46
SYS
So, you know, you have a lot of support on this side, so you can do it. Okay. Remember, you have to support on the QI. You can support it. You have a lot of pressure. You have to support it. I'm done.
09:47
SYS
You asked... You ready? The record? Di plugins Double Colours? Yes, I want the formal title. It will be 8. I hope you can sign up for this FCO with less ACRAeren Production good-bye. Dear Tricks. Amen.
09:47
SYS
I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. I'm not sure if you can see it. We'll see you in the next one.
09:48
SYS
Good. That's not the end. Is it? I'm going to buy it. You can buy it for me. I'm going. Good. No one is going to buy it. I'll buy it. I'll buy it. Where is the test? all of it.
09:48
SYS
I mean, I don't know how to do it. I mean, I don't know how to do it. I mean, I don't know how to do it. Good. I don't know how to do it. I don't know how to do it. And then.
09:49
SYS
From 36, we've heard from you, from the left, from the right. They... They gave us the same thing. For the other one, for the other one, for the other one. Okay. That's how it looks like.
09:49
SYS
And I do this as a key rate, as you can see. Okay, thank you. And over here, in progress, Nikki. I will show you the first piece of file-shaving application for the case of Google.
09:50
SYS
I have to say that the circle of the people in the world is a single day, so I started to argue. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same. I have to say that I have to be the same.
09:50
SYS
That's what I'm talking about. That's what I'm talking about. I'm talking about the fact that I'm not a professional, but I'm not a professional. I'm talking about the fact that I'm not a professional, but I'm a professional. I'm talking about the fact that I'm a professional. That's what I'm talking about. I'm still there.
09:51
SYS
I mean, the KMOS is very difficult to work with the team. It's a 2-sub-tory deployment. The first step is to make the decision on the team.
09:51
SYS
I think it's a good idea. Why do you have to go to the point? I don't know. And I think that's it. Because the next question is about the day of the day, but it's a good result. And it's not a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. See you soon. Bye.
09:51
SYS
I don't know how many years I've been reading the books, but I'm not sure if you're going to read it, but I think the most important thing is that the most important thing is that the three of us are going to try to do it in the next 10 years. And we are going to try to find out what we are going to do in the next 10 years.
09:52
SYS
So, the last maximum moment in which you can't reach the point, but not in any of these places. You can see that the coastline is not very far from the coastline. It's just a very long journey. Jason. I don't know, but I don't know what to do. I mean, I think that the... the direction of the sea is the reality. The word is the word, the word is the word, the word is the word, the word is the word.
09:52
SYS
And I think we will also have a better future. So, this is a mini test. I don't know where it is. So, this is a very important story. Thank you.
09:53
SYS
There are many ways to develop the idea of the post-cum model. Because we don't have to do it. We have to do it. I don't have the idea of the idea of the post-cum. If you are a long post-cum, you have to do it. And you have to do it. That's why. I'm an American. I have my two-part post-cum. for two.
09:53
SYS
same. I'm going to talk about the next episode.
09:54
SYS
But what is the most important thing in the world? We don't have a problem. We can say that we are not in the world.
09:54
SYS
I think it's a good idea. And I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. I think it's a good idea. the most important thing in the world.
09:55
SYS
I don't have a lot of resources.
09:55
SYS
I'm not sure if you can see it. I'm not sure if you can see it.
09:56
SYS
I think that the budget is not enough, but I'll just go around. I think that's a good idea. That's a good idea. The weight is almost like a low weight. The weight is 4.5 kg.
09:56
SYS
I think it's a good idea. I think it's a good idea.
09:57
SYS
I'm sorry.
09:57
SYS
I'm going to put it in the middle of the first private, because it's not on the head. I'll put it on the head. And then I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. I'll put it on the head. could check both of the arguments very
09:58
SYS
Well, I'm going to give you a few words. How I would like to be able to do something. I'm going to give you a few more questions. How I would like to be able to do something that I would like to do. I would like to give you a few more questions. I would like to give you a few more questions. So, this is a great question.
09:59
SYS
I don't know if you can see the changes in the future. I don't know if you can see the changes in the future. You can see the changes in the future. Yes. I think that's the first time I've seen the changes in the future.
09:59
SYS
I mean, we want to be a game changer, some problems with the idea of the game of the game. We have something to do with the game changer. We want to make a filter for the first time in the world. We want to make a game changer, but it's not a good idea. We want to make a filter for the game changer.
10:00
SYS
I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I think it's a bit different. I don't know if it's a bit different. I think it's a bit different. I think it's a bit different. and the community, as a support, we wish you a good day.
10:00
SYS
We are here in the beginning of the game. And now, we can start the game with the first one, the first one, the third one, the second one. The third one, the fourth one. The fourth one. We can start the game. The first one, the first one, the first one.
10:01
SYS
because it's near.
10:01
SYS
Yo, y'all, no connect, no connect.
10:01
SYS
How did you see that? How did you see that? We don't know anything. Well, I don't know anything. I was talking about the article, I didn't know the comments, but it was a theory. I was going to say, well, it was a mistake. It started like this. Yes, it was a new topic, I'm sure you'll find a link in the description. And I'll check the results.
10:02
SYS
I don't know. I don't know. I'm a person who can help you. Maybe I'll be able to use this. Coinir client. Coinir client. Coinir client. I'm a person who can help you. Coinir client. What's your client? I'm a person who can help you. No, it's not a client. I don't know. I don't know. I don't know. Coinir client.
10:02
SYS
What if you don't have any technical problems? Or are you going to be in the middle of the day? And that's the reason. And that's the reason. That's the reason for the first day or the second day. That's the reason for the first time. Then we know, the data is the format. That's the reason for the first day or the second day.
10:03
SYS
I think that's the way to connect the content of the content. It's not the first time we've been in the world. What's your name? I mean, I'm a manager of mine. No, I don't know. I'll see you next time. I'll see you next time.
10:03
SYS
Well, the question was about the complicated for 5% of the users, the complicated, the advanced VDC5. What is the point of the new version? The features of the 5-year user and the user, I think, are you sure? It's not a problem. You know, this is not a matter of fact. It's a pre-incialization of the chat. It's a matter of fact. It's a kind of a topic. But we can't say that, we can't go further.
10:04
SYS
And I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. And that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. I think that's the most important thing. Oh, I'm sorry.
10:04
SYS
And I'm not sure what you're going to use. Yes, because at least it's a very interesting thing. I'm not sure if you can see it. I don't know if you're going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. I'm going to be in the same room. This is the color.
10:04
SYS
What do you think about this?
10:05
SYS
Good. You're the best in the world. Okay. Okay. I'll show you how to print. So, I know that this is the case in which I'm working on the program. I'm going to print it. And I'm going to print it.
10:05
SYS
I'm not a British. Yes. I'm not sure what it is. I'm not a British.
10:06
SYS
review the situation in the next sentence. Good. And the two styles can be understood. Well, I'm not sure if I'm going to be able to do it. 5, that's all. I'm sure we're not going to be able to do it. I'm going to be able to do it. and poke a motion. See you in the next video.
10:06
SYS
Back for me because she was...
10:06
SYS
Second problem is that no problem, you have to be able to do something, fix it and you have to do something, you have to do it.
10:07
SYS
I think it's a bit more like a little bit. And it's still a bit more like a little bit more. So I think it's a bit more like a little bit more. Like in the second one, I don't know. Maybe it's a bit more like a little bit more. on the whole video.
10:08
SYS
And then?
10:08
SYS
I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I think that's the reason why I'm doing this. I'm going to go to the next one.
10:09
SYS
What is your main plan on my dinner just to move this man out of the seahab? What is your main plan for the seahab? Considering you red, thank you. Thank you very much. Thank you.
10:09
SYS
I'm almost there.
10:10
SYS
And where are the other two interplayers? I have a processing system.
10:10
SYS
I think it's a good idea. I think it's a good idea.
10:11
SYS
I've got a lot of stuff to do with the content. I don't know what I'm talking about. I've got a lot of stuff to do with the project. I'm talking about the project. I'm talking about the project. I'm talking about the project. So, let's talk to the user.
10:12
MIC
I'm gonna use a little bit of water to prevent the water from sticking to the water.
10:12
SYS
That's the way to analyze the neural network. That's how we did it. That's how we did it. That's how we did it. That's how we did it.
10:12
SYS
I'm sorry. Thank you very much. I'm sorry.
10:13
SYS
And then he died, he died, he died. Yes, I remember the rules of the rules of the war. And I remember the rules of the war, that is, that is, that is, that is, the current situation is happening. And I remember the rules of the war.
10:13
SYS
But it's been good. Yeah, it was good. Yeah, it was good. So, I'd aim or select them. And then, so, I'll be able to...
10:14
SYS
didn't make some sense....
|
NULL
|
NULL
|
NULL
|
NULL
|
|
32065
|
1240
|
43
|
2026-05-13T09:04:58.844684+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-13/1778 /Users/lukas/.screenpipe/data/data/2026-05-13/1778663098844_m1.jpg...
|
iTerm2
|
-zsh
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
+VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:6 +VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:63> (( 1 ))
+VCS_INFO_quilt:86> (( 1 ))
+VCS_INFO_quilt:92> emulate -L zsh
+VCS_INFO_quilt:93> setopt extendedglob
+VCS_INFO_quilt:94> local mode=standalone
+VCS_INFO_quilt:95> local patches pc qstring root
+VCS_INFO_quilt:96> local -i ret
+VCS_INFO_quilt:97> local context
+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env
+VCS_INFO_quilt:99> local -A hook_com
+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all-
+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt
+VCS_INFO_quilt:102> return 1
+vcs_info:82> VCS_INFO_set --nvcs
+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset
+VCS_INFO_set:8> local -i i j
+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]
+VCS_INFO_set:11> [[ '' == -preinit- ]]
+VCS_INFO_set:12> i=0
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='
+VCS_INFO_set:12> i=1
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='
+VCS_INFO_set:15> VCS_INFO_nvcsformats
+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit
+VCS_INFO_nvcsformats:6> local c v rr
+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]
+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs
+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))
+VCS_INFO_nvcsformats:15> return 0
+VCS_INFO_set:16> [[ '' != -preinit- ]]
+VCS_INFO_set:16> VCS_INFO_hook no-vcs
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=no-vcs
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs
+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+VCS_INFO_set:19> (( 0 - 1 < 0 ))
+VCS_INFO_set:19> return 0
+vcs_info:83> return 0
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts $ cd lib
+-zsh:65> cd lib
+precmd:0> vcs_info
+vcs_info:1> emulate -L zsh
+vcs_info:2> setopt extendedglob NO_warn_create_global
+vcs_info:4> [[ -r . ]]
+vcs_info:6> local pat
+vcs_info:7> local -i found retval
+vcs_info:8> local -a enabled disabled dps
+vcs_info:9> local usercontext vcs rrn quiltmode
+vcs_info:10> local -x LC_MESSAGES
+vcs_info:11> local -i maxexports
+vcs_info:12> local -a msgs
+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data
+vcs_info:21> LC_MESSAGES=C
+vcs_info:22> [[ -n '' ]]
+vcs_info:27> vcs=-init-
+vcs_info:27> rrn=-all-
+vcs_info:27> quiltmode=addon
+vcs_info:28> usercontext=default
+vcs_info:30> VCS_INFO_hook start-up
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=start-up
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up
+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+vcs_info:31> retval=0
+vcs_info:32> (( retval == 1 ))
+vcs_info:34> (( retval == 2 ))
+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled
+vcs_info:43> (( 0 == 0 ))
+vcs_info:43> enabled=( all )
+vcs_info:45> [[ -n '' ]]
+vcs_info:50> [[ -n all ]]
+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla )
+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled
+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps
+vcs_info:65> VCS_INFO_maxexports
+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset
+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports
+VCS_INFO_maxexports:7> maxexports=2
+VCS_INFO_maxexports:8> [[ 2 != <-> ]]
+VCS_INFO_maxexports:8> (( maxexports < 1 ))
+VCS_INFO_maxexports:13> return 0
+vcs_info:67> (( found = 0 ))
+vcs_info:68> vcs=bzr
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr
+vcs_info:77> VCS_INFO_detect_bzr
+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]
+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case bzr (/*)
+VCS_INFO_check_com:7> case bzr (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_bzr:9> return 1
+vcs_info:68> vcs=cdv
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv
+vcs_info:77> VCS_INFO_detect_cdv
+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cdv (/*)
+VCS_INFO_check_com:7> case cdv (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cdv:9> return 1
+vcs_info:68> vcs=cvs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs
+vcs_info:77> VCS_INFO_detect_cvs
+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cvs (/*)
+VCS_INFO_check_com:7> case cvs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cvs:9> return 1
+vcs_info:68> vcs=darcs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs
+vcs_info:77> VCS_INFO_detect_darcs
+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case darcs (/*)
+VCS_INFO_check_com:7> case darcs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_darcs:9> return 1
+vcs_info:68> vcs=fossil
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil
+vcs_info:77> VCS_INFO_detect_fossil
+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]
+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case fossil (/*)
+VCS_INFO_check_com:7> case fossil (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_fossil:9> return 1
+vcs_info:68> vcs=git
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git
+vcs_info:77> VCS_INFO_detect_git
+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_git:7> [[ '' == --flavours ]]
+VCS_INFO_detect_git:9> VCS_INFO_check_com git
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case git (/*)
+VCS_INFO_check_com:7> case git (*)
+VCS_INFO_check_com:12> (( 1 ))
+VCS_INFO_check_com:12> return 0
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=''
+VCS_INFO_detect_git:15> return 1
+vcs_info:68> vcs=hg
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg
+vcs_info:77> VCS_INFO_detect_hg
+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]
+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case hg (/*)
+VCS_INFO_check_com:7> case hg (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_hg:9> return 1
+vcs_info:68> vcs=mtn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn
+vcs_info:77> VCS_INFO_detect_mtn
+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case mtn (/*)
+VCS_INFO_check_com:7> case mtn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_mtn:9> return 1
+vcs_info:68> vcs=p4
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4
+vcs_info:77> VCS_INFO_detect_p4
+VCS_INFO_detect_p4:1> local serverport p4where
+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server
+VCS_INFO_detect_p4:23> [[ -n '' ]]
+VCS_INFO_detect_p4:23> return 1
+vcs_info:68> vcs=svk
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk
+vcs_info:77> VCS_INFO_detect_svk
+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob
+VCS_INFO_detect_svk:17> local -i fhash
+VCS_INFO_detect_svk:18> fhash=0
+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svk (/*)
+VCS_INFO_check_com:7> case svk (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svk:20> return 1
+vcs_info:68> vcs=svn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn
+vcs_info:77> VCS_INFO_detect_svn
+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svn (/*)
+VCS_INFO_check_com:7> case svn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svn:9> return 1
+vcs_info:68> vcs=tla
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla
+vcs_info:77> VCS_INFO_detect_tla
+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]
+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case tla (/*)
+VCS_INFO_check_com:7> case tla (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_tla:9> return 1
+vcs_info:80> (( found == 0 ))
+vcs_info:81> vcs=-quilt-
+vcs_info:81> quiltmode=standalone
+vcs_info:82> VCS_INFO_quilt standalone
+VCS_INFO_quilt:3> (( 1 ))
+VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:63> (( 1 ))
+VCS_INFO_quilt:86> (( 1 ))
+VCS_INFO_quilt:92> emulate -L zsh
+VCS_INFO_quilt:93> setopt extendedglob
+VCS_INFO_quilt:94> local mode=standalone
+VCS_INFO_quilt:95> local patches pc qstring root
+VCS_INFO_quilt:96> local -i ret
+VCS_INFO_quilt:97> local context
+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env
+VCS_INFO_quilt:99> local -A hook_com
+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all-
+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt
UW PICO 5.09 New Buffer
[ Read 297 lines ]
^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos
^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell
+VCS_INFO_quilt:102> return 1
+vcs_info:82> VCS_INFO_set --nvcs
+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset
+VCS_INFO_set:8> local -i i j
+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]
+VCS_INFO_set:11> [[ '' == -preinit- ]]
+VCS_INFO_set:12> i=0
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='
+VCS_INFO_set:12> i=1
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='
+VCS_INFO_set:15> VCS_INFO_nvcsformats
+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit
+VCS_INFO_nvcsformats:6> local c v rr
+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]
+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs
+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))
+VCS_INFO_nvcsformats:15> return 0
+VCS_INFO_set:16> [[ '' != -preinit- ]]
+VCS_INFO_set:16> VCS_INFO_hook no-vcs
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=no-vcs
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs
+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+VCS_INFO_set:19> (( 0 - 1 < 0 ))
+VCS_INFO_set:19> return 0
+vcs_info:83> return 0
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh
+-zsh:66> nano screenpipe_sync_db.sh
+precmd:0> vcs_info
+vcs_info:1> emulate -L zsh
+vcs_info:2> setopt extendedglob NO_warn_create_global
+vcs_info:4> [[ -r . ]]
+vcs_info:6> local pat
+vcs_info:7> local -i found retval
+vcs_info:8> local -a enabled disabled dps
+vcs_info:9> local usercontext vcs rrn quiltmode
+vcs_info:10> local -x LC_MESSAGES
+vcs_info:11> local -i maxexports
+vcs_info:12> local -a msgs
+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data
+vcs_info:21> LC_MESSAGES=C
+vcs_info:22> [[ -n '' ]]
+vcs_info:27> vcs=-init-
+vcs_info:27> rrn=-all-
+vcs_info:27> quiltmode=addon
+vcs_info:28> usercontext=default
+vcs_info:30> VCS_INFO_hook start-up
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=start-up
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up
+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+vcs_info:31> retval=0
+vcs_info:32> (( retval == 1 ))
+vcs_info:34> (( retval == 2 ))
+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled
+vcs_info:43> (( 0 == 0 ))
+vcs_info:43> enabled=( all )
+vcs_info:45> [[ -n '' ]]
+vcs_info:50> [[ -n all ]]
+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla )
+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled
+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps
+vcs_info:65> VCS_INFO_maxexports
+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset
+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports
+VCS_INFO_maxexports:7> maxexports=2
+VCS_INFO_maxexports:8> [[ 2 != <-> ]]
+VCS_INFO_maxexports:8> (( maxexports < 1 ))
+VCS_INFO_maxexports:13> return 0
+vcs_info:67> (( found = 0 ))
+vcs_info:68> vcs=bzr
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr
+vcs_info:77> VCS_INFO_detect_bzr
+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]
+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case bzr (/*)
+VCS_INFO_check_com:7> case bzr (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_bzr:9> return 1
+vcs_info:68> vcs=cdv
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv
+vcs_info:77> VCS_INFO_detect_cdv
+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cdv (/*)
+VCS_INFO_check_com:7> case cdv (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cdv:9> return 1
+vcs_info:68> vcs=cvs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs
+vcs_info:77> VCS_INFO_detect_cvs
+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cvs (/*)
+VCS_INFO_check_com:7> case cvs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cvs:9> return 1
+vcs_info:68> vcs=darcs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs
+vcs_info:77> VCS_INFO_detect_darcs
+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case darcs (/*)
+VCS_INFO_check_com:7> case darcs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_darcs:9> return 1
+vcs_info:68> vcs=fossil
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil
+vcs_info:77> VCS_INFO_detect_fossil
+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]
+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case fossil (/*)
+VCS_INFO_check_com:7> case fossil (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_fossil:9> return 1
+vcs_info:68> vcs=git
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git
+vcs_info:77> VCS_INFO_detect_git
+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_git:7> [[ '' == --flavours ]]
+VCS_INFO_detect_git:9> VCS_INFO_check_com git
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case git (/*)
+VCS_INFO_check_com:7> case git (*)
+VCS_INFO_check_com:12> (( 1 ))
+VCS_INFO_check_com:12> return 0
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=''
+VCS_INFO_detect_git:15> return 1
+vcs_info:68> vcs=hg
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg
+vcs_info:77> VCS_INFO_detect_hg
+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]
+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case hg (/*)
+VCS_INFO_check_com:7> case hg (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_hg:9> return 1
+vcs_info:68> vcs=mtn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn
+vcs_info:77> VCS_INFO_detect_mtn
+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case mtn (/*)
+VCS_INFO_check_com:7> case mtn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_mtn:9> return 1
+vcs_info:68> vcs=p4
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4
+vcs_info:77> VCS_INFO_detect_p4
+VCS_INFO_detect_p4:1> local serverport p4where
+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server
+VCS_INFO_detect_p4:23> [[ -n '' ]]
+VCS_INFO_detect_p4:23> return 1
+vcs_info:68> vcs=svk
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk
+vcs_info:77> VCS_INFO_detect_svk
+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob
+VCS_INFO_detect_svk:17> local -i fhash
+VCS_INFO_detect_svk:18> fhash=0
+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svk (/*)
+VCS_INFO_check_com:7> case svk (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svk:20> return 1
+vcs_info:68> vcs=svn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn
+vcs_info:77> VCS_INFO_detect_svn
+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svn (/*)
+VCS_INFO_check_com:7> case svn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svn:9> return 1
+vcs_info:68> vcs=tla
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla
+vcs_info:77> VCS_INFO_detect_tla
+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]
+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case tla (/*)
+VCS_INFO_check_com:7> case tla (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_tla:9> return 1
+vcs_info:80> (( found == 0 ))
+vcs_info:81> vcs=-quilt-
+vcs_info:81> quiltmode=standalone
+vcs_info:82> VCS_INFO_quilt standalone
+VCS_INFO_quilt:3> (( 1 ))
+VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:63> (( 1 ))
+VCS_INFO_quilt:86> (( 1 ))
+VCS_INFO_quilt:92> emulate -L zsh
+VCS_INFO_quilt:93> setopt extendedglob
+VCS_INFO_quilt:94> local mode=standalone
+VCS_INFO_quilt:95> local patches pc qstring root
+VCS_INFO_quilt:96> local -i ret
+VCS_INFO_quilt:97> local context
+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env
+VCS_INFO_quilt:99> local -A hook_com
+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all-
+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt
+VCS_INFO_quilt:102> return 1
+vcs_info:82> VCS_INFO_set --nvcs
+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset
+VCS_INFO_set:8> local -i i j
+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]
+VCS_INFO_set:11> [[ '' == -preinit- ]]
+VCS_INFO_set:12> i=0
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='
+VCS_INFO_set:12> i=1
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='
+VCS_INFO_set:15> VCS_INFO_nvcsformats
+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit
+VCS_INFO_nvcsformats:6> local c v rr
+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]
+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs
+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))
+VCS_INFO_nvcsformats:15> return 0
+VCS_INFO_set:16> [[ '' != -preinit- ]]
+VCS_INFO_set:16> VCS_INFO_hook no-vcs
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=no-vcs
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs
+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+VCS_INFO_set:19> (( 0 - 1 < 0 ))
+VCS_INFO_set:19> return 0
+vcs_info:83> return 0
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ set +x
+-zsh:67> set +x
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-11
[2026-05-13 11:40:52] ========================================
[2026-05-13 11:40:52] Screenpipe sync starting for: 2026-05-11
[2026-05-13 11:40:52] ========================================
[+00m00s] ▶ Preflight checks
Source DB: OK (4.5G)
NAS mount: OK /Volumes/screenpipe
[2026-05-13 11:40:53] Date 2026-05-11 already has 6857 frames in archive — skipping DB sync
Data dir: OK (283 files, 318M)
[+00m01s] ▶ Copying data folder for 2026-05-11
rsync 2026-05-11/ → NAS ✓ 0m01s (283 files, 318M)
[+00m02s] ▶ Copying audio files for 2026-05-11
rsync audio files → NAS skipped (no audio files for date)
[+00m02s] ▶ Copying screenpipe logs for 2026-05-11
rsync logs → NAS ✓ 1 file(s), 520K
[2026-05-13 11:40:54] Archive DB size: 1.3G
[2026-05-13 11:40:54] Total time: 0m2s
[2026-05-13 11:40:54] Sync complete for 2026-05-11
[2026-05-13 11:40:54] ========================================
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12
[2026-05-13 11:41:34] ========================================
UW PICO 5.09 New Buffer
[ Read 297 lines ]
^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos
^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell
UW PICO 5.09 File: screenpipe_sync_db.sh
Pico Help Text
Pico is designed to be a simple, easy-to-use text editor with a
layout very similar to the Alpine mailer. The status line at the
top of the display shows pico's version, the current file being
edited and whether or not there are outstanding modifications
that have not been saved. The third line from the bottom is used
to report informational messages and for additional command input.
The bottom two lines list the available editing commands.
Each character typed is automatically inserted into the buffer
at the current cursor position. Editing commands and cursor
movement (besides arrow keys) are given to pico by typing
special control-key sequences. A caret, '^', is used to denote
the control key, sometimes marked "CTRL", so the CTRL-q key
combination is written as ^Q.
The following functions are available in pico (where applicable,
corresponding function key commands are in parentheses).
^G (F1) Display this help text.
^F move Forward a character.
^B move Backward a character.
^P move to the Previous line.
^N move to the Next line.
^A move to the beginning of the current line.
^E move to the End of the current line.
^V (F8) move forward a page of text.
^Y (F7) move backward a page of text.
^W (F6) Search for (where is) text, neglecting case.
^L Refresh the display.
^D Delete the character at the cursor position.
^^ Mark cursor position as beginning of selected text.
Note: Setting mark when already set unselects text.
^K (F9) Cut selected text (displayed in inverse characters).
Note: The selected text's boundary on the cursor side
ends at the left edge of the cursor. So, with
^X Exit Help ^V Next Pg
[2026-05-13 11:41:34] Screenpipe sync starting for: 2026-05-12
[2026-05-13 11:41:34] ========================================
[+00m00s] ▶ Preflight checks
Source DB: OK (4.5G)
NAS mount: OK /Volumes/screenpipe
Archive DB: exists (1.3G)
Data dir: OK (220 files, 303M)
[+00m00s] ▶ Counting source rows for 2026-05-12
frames: 7274
elements: 853406
ui_events: 7044
ocr_text: 2306
meetings: 3
audio_chunks: 1113
audio_transcriptions: 319
[+00m01s] ▶ Initialising tables, indexes, FTS
creating tables ✓ 0m00s
creating indexes ✓ 0m00s
creating FTS tables ✓ 0m00s
[+00m01s] ▶ Syncing vision data for 2026-05-12
video_chunks ✓ 0m01s
frames (7274 rows) ✓ 1m05s
ocr_text (2306 rows) ✓ 0m49s
ui_events (7044 rows) ✓ 0m01s
elements (853406 rows) ✓ 0m46s
meetings (3 rows) ✓ 0m00s
[+02m43s] ▶ Syncing audio data for 2026-05-12
audio_chunks (1113 rows) ✓ 0m00s
audio_transcriptions (319 rows) ✓ 0m01s
[+02m44s] ▶ Updating FTS indexes
elements_fts ✓ 0m50s
frames_fts ✓ 1m13s
ui_events_fts ✓ 0m01s
audio_transcriptions_fts ✓ 0m01s
[+04m49s] ▶ Verifying DB
/Users/lukas/.screenpipe/scripts/lib/screenpipe_sync_db.sh: line 220: target_DATE: unbound variable
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12
[2026-05-13 11:53:26] ========================================
[2026-05-13 11:53:26] Screenpipe sync starting for: 2026-05-12
[2026-05-13 11:53:26] ========================================
[+00m00s] ▶ Preflight checks
Source DB: OK (4.6G)
NAS mount: OK /Volumes/screenpipe
Archive DB: will be created
Data dir: OK (220 files, 303M)
[+00m00s] ▶ Counting source rows for 2026-05-12
frames: 7274
elements: 853406
ui_events: 7044
ocr_text: 2306
meetings: 3
audio_chunks: 1113
audio_transcriptions: 319
[+00m00s] ▶ Initialising tables, indexes, FTS
creating tables ✓ 0m00s
creating indexes ✓ 0m01s
creating FTS tables ✓ 0m00s
[+00m02s] ▶ Syncing vision data for 2026-05-12
video_chunks ✓ 0m00s
frames (7274 rows) ✓ 1m03s
ocr_text (2306 rows) ✓ 0m48s
ui_events (7044 rows) ✓ 0m01s
elements (853406 rows) ✓ 0m46s
meetings (3 rows) ✓ 0m00s
[+02m40s] ▶ Syncing audio data for 2026-05-12
audio_chunks (1113 rows) ✓ 0m00s
audio_transcriptions (319 rows) ✓ 0m00s
[+02m40s] ▶ Updating FTS indexes
elements_fts ✓ 0m32s
frames_fts ✓ 1m08s
ui_events_fts ✓ 0m00s
audio_transcriptions_fts ✓ 0m00s
[+04m20s] ▶ Verifying DB
frames: 7274 / 7274 ✓
elements: 853406 / 853406 ✓
ui_events: 7044 / 7044 ✓
ocr_text: 2306 / 2306 ✓
meetings: 3 / 3 ✓
audio_chunks: 1113 / 1113 ✓
audio_transcriptions: 319 / 319 ✓
[+04m29s] ▶ Copying data folder for 2026-05-12
rsync 2026-05-12/ → NAS ✓ 0m20s (220 files, 295M)
[+04m49s] ▶ Copying audio files for 2026-05-12
rsync audio files → NAS skipped (no audio files for date)
[+04m49s] ▶ Copying screenpipe logs for 2026-05-12
rsync logs → NAS ✓ 1 file(s), 304K
[2026-05-13 11:58:15] Archive DB size: 763M
[2026-05-13 11:58:15] Total time: 4m49s
[2026-05-13 11:58:15] Sync complete for 2026-05-12
[2026-05-13 11:58:15] ========================================
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $
DOCKER
Close Tab
DEV (-zsh)
Close Tab
APP (-zsh)
Close Tab
-zsh
Close Tab
screenpipe"
Close Tab
⌥⌘1
-zsh...
|
[{"role":"AXTextArea","text [{"role":"AXTextArea","text":"+VCS_INFO_quilt:24> (( 1 ))\n+VCS_INFO_quilt:63> (( 1 ))\n+VCS_INFO_quilt:86> (( 1 ))\n+VCS_INFO_quilt:92> emulate -L zsh\n+VCS_INFO_quilt:93> setopt extendedglob\n+VCS_INFO_quilt:94> local mode=standalone\n+VCS_INFO_quilt:95> local patches pc qstring root\n+VCS_INFO_quilt:96> local -i ret\n+VCS_INFO_quilt:97> local context\n+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env\n+VCS_INFO_quilt:99> local -A hook_com\n+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all- \n+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt\n+VCS_INFO_quilt:102> return 1\n+vcs_info:82> VCS_INFO_set --nvcs\n+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset\n+VCS_INFO_set:8> local -i i j\n+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]\n+VCS_INFO_set:11> [[ '' == -preinit- ]]\n+VCS_INFO_set:12> i=0\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='\n+VCS_INFO_set:12> i=1\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='\n+VCS_INFO_set:15> VCS_INFO_nvcsformats\n+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit\n+VCS_INFO_nvcsformats:6> local c v rr\n+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]\n+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs\n+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))\n+VCS_INFO_nvcsformats:15> return 0\n+VCS_INFO_set:16> [[ '' != -preinit- ]]\n+VCS_INFO_set:16> VCS_INFO_hook no-vcs\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=no-vcs \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+VCS_INFO_set:19> (( 0 - 1 < 0 ))\n+VCS_INFO_set:19> return 0\n+vcs_info:83> return 0\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts $ cd lib \n+-zsh:65> cd lib\n+precmd:0> vcs_info \n+vcs_info:1> emulate -L zsh\n+vcs_info:2> setopt extendedglob NO_warn_create_global\n+vcs_info:4> [[ -r . ]]\n+vcs_info:6> local pat\n+vcs_info:7> local -i found retval\n+vcs_info:8> local -a enabled disabled dps\n+vcs_info:9> local usercontext vcs rrn quiltmode\n+vcs_info:10> local -x LC_MESSAGES\n+vcs_info:11> local -i maxexports\n+vcs_info:12> local -a msgs\n+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data\n+vcs_info:21> LC_MESSAGES=C \n+vcs_info:22> [[ -n '' ]]\n+vcs_info:27> vcs=-init- \n+vcs_info:27> rrn=-all- \n+vcs_info:27> quiltmode=addon \n+vcs_info:28> usercontext=default \n+vcs_info:30> VCS_INFO_hook start-up\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=start-up \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+vcs_info:31> retval=0 \n+vcs_info:32> (( retval == 1 ))\n+vcs_info:34> (( retval == 2 ))\n+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled\n+vcs_info:43> (( 0 == 0 ))\n+vcs_info:43> enabled=( all ) \n+vcs_info:45> [[ -n '' ]]\n+vcs_info:50> [[ -n all ]]\n+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla ) \n+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled\n+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps\n+vcs_info:65> VCS_INFO_maxexports\n+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset\n+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports\n+VCS_INFO_maxexports:7> maxexports=2 \n+VCS_INFO_maxexports:8> [[ 2 != <-> ]]\n+VCS_INFO_maxexports:8> (( maxexports < 1 ))\n+VCS_INFO_maxexports:13> return 0\n+vcs_info:67> (( found = 0 ))\n+vcs_info:68> vcs=bzr\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr \n+vcs_info:77> VCS_INFO_detect_bzr\n+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case bzr (/*)\n+VCS_INFO_check_com:7> case bzr (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_bzr:9> return 1\n+vcs_info:68> vcs=cdv\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv \n+vcs_info:77> VCS_INFO_detect_cdv\n+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cdv (/*)\n+VCS_INFO_check_com:7> case cdv (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cdv:9> return 1\n+vcs_info:68> vcs=cvs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs \n+vcs_info:77> VCS_INFO_detect_cvs\n+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cvs (/*)\n+VCS_INFO_check_com:7> case cvs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cvs:9> return 1\n+vcs_info:68> vcs=darcs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs \n+vcs_info:77> VCS_INFO_detect_darcs\n+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case darcs (/*)\n+VCS_INFO_check_com:7> case darcs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_darcs:9> return 1\n+vcs_info:68> vcs=fossil\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil \n+vcs_info:77> VCS_INFO_detect_fossil\n+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case fossil (/*)\n+VCS_INFO_check_com:7> case fossil (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_fossil:9> return 1\n+vcs_info:68> vcs=git\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git \n+vcs_info:77> VCS_INFO_detect_git\n+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_git:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_git:9> VCS_INFO_check_com git\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case git (/*)\n+VCS_INFO_check_com:7> case git (*)\n+VCS_INFO_check_com:12> (( 1 ))\n+VCS_INFO_check_com:12> return 0\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]='' \n+VCS_INFO_detect_git:15> return 1\n+vcs_info:68> vcs=hg\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg \n+vcs_info:77> VCS_INFO_detect_hg\n+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case hg (/*)\n+VCS_INFO_check_com:7> case hg (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_hg:9> return 1\n+vcs_info:68> vcs=mtn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn \n+vcs_info:77> VCS_INFO_detect_mtn\n+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case mtn (/*)\n+VCS_INFO_check_com:7> case mtn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_mtn:9> return 1\n+vcs_info:68> vcs=p4\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4 \n+vcs_info:77> VCS_INFO_detect_p4\n+VCS_INFO_detect_p4:1> local serverport p4where\n+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server\n+VCS_INFO_detect_p4:23> [[ -n '' ]]\n+VCS_INFO_detect_p4:23> return 1\n+vcs_info:68> vcs=svk\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk \n+vcs_info:77> VCS_INFO_detect_svk\n+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob\n+VCS_INFO_detect_svk:17> local -i fhash\n+VCS_INFO_detect_svk:18> fhash=0 \n+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svk (/*)\n+VCS_INFO_check_com:7> case svk (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svk:20> return 1\n+vcs_info:68> vcs=svn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn \n+vcs_info:77> VCS_INFO_detect_svn\n+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svn (/*)\n+VCS_INFO_check_com:7> case svn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svn:9> return 1\n+vcs_info:68> vcs=tla\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla \n+vcs_info:77> VCS_INFO_detect_tla\n+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case tla (/*)\n+VCS_INFO_check_com:7> case tla (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_tla:9> return 1\n+vcs_info:80> (( found == 0 ))\n+vcs_info:81> vcs=-quilt- \n+vcs_info:81> quiltmode=standalone \n+vcs_info:82> VCS_INFO_quilt standalone\n+VCS_INFO_quilt:3> (( 1 ))\n+VCS_INFO_quilt:24> (( 1 ))\n+VCS_INFO_quilt:63> (( 1 ))\n+VCS_INFO_quilt:86> (( 1 ))\n+VCS_INFO_quilt:92> emulate -L zsh\n+VCS_INFO_quilt:93> setopt extendedglob\n+VCS_INFO_quilt:94> local mode=standalone\n+VCS_INFO_quilt:95> local patches pc qstring root\n+VCS_INFO_quilt:96> local -i ret\n+VCS_INFO_quilt:97> local context\n+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env\n+VCS_INFO_quilt:99> local -A hook_com\n+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all- \n+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt\n UW PICO 5.09 New Buffer \n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n [ Read 297 lines ] \n^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos \n^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell \n+VCS_INFO_quilt:102> return 1\n+vcs_info:82> VCS_INFO_set --nvcs\n+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset\n+VCS_INFO_set:8> local -i i j\n+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]\n+VCS_INFO_set:11> [[ '' == -preinit- ]]\n+VCS_INFO_set:12> i=0\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='\n+VCS_INFO_set:12> i=1\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='\n+VCS_INFO_set:15> VCS_INFO_nvcsformats\n+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit\n+VCS_INFO_nvcsformats:6> local c v rr\n+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]\n+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs\n+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))\n+VCS_INFO_nvcsformats:15> return 0\n+VCS_INFO_set:16> [[ '' != -preinit- ]]\n+VCS_INFO_set:16> VCS_INFO_hook no-vcs\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=no-vcs \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+VCS_INFO_set:19> (( 0 - 1 < 0 ))\n+VCS_INFO_set:19> return 0\n+vcs_info:83> return 0\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh\n+-zsh:66> nano screenpipe_sync_db.sh\n+precmd:0> vcs_info \n+vcs_info:1> emulate -L zsh\n+vcs_info:2> setopt extendedglob NO_warn_create_global\n+vcs_info:4> [[ -r . ]]\n+vcs_info:6> local pat\n+vcs_info:7> local -i found retval\n+vcs_info:8> local -a enabled disabled dps\n+vcs_info:9> local usercontext vcs rrn quiltmode\n+vcs_info:10> local -x LC_MESSAGES\n+vcs_info:11> local -i maxexports\n+vcs_info:12> local -a msgs\n+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data\n+vcs_info:21> LC_MESSAGES=C \n+vcs_info:22> [[ -n '' ]]\n+vcs_info:27> vcs=-init- \n+vcs_info:27> rrn=-all- \n+vcs_info:27> quiltmode=addon \n+vcs_info:28> usercontext=default \n+vcs_info:30> VCS_INFO_hook start-up\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=start-up \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+vcs_info:31> retval=0 \n+vcs_info:32> (( retval == 1 ))\n+vcs_info:34> (( retval == 2 ))\n+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled\n+vcs_info:43> (( 0 == 0 ))\n+vcs_info:43> enabled=( all ) \n+vcs_info:45> [[ -n '' ]]\n+vcs_info:50> [[ -n all ]]\n+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla ) \n+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled\n+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps\n+vcs_info:65> VCS_INFO_maxexports\n+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset\n+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports\n+VCS_INFO_maxexports:7> maxexports=2 \n+VCS_INFO_maxexports:8> [[ 2 != <-> ]]\n+VCS_INFO_maxexports:8> (( maxexports < 1 ))\n+VCS_INFO_maxexports:13> return 0\n+vcs_info:67> (( found = 0 ))\n+vcs_info:68> vcs=bzr\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr \n+vcs_info:77> VCS_INFO_detect_bzr\n+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case bzr (/*)\n+VCS_INFO_check_com:7> case bzr (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_bzr:9> return 1\n+vcs_info:68> vcs=cdv\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv \n+vcs_info:77> VCS_INFO_detect_cdv\n+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cdv (/*)\n+VCS_INFO_check_com:7> case cdv (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cdv:9> return 1\n+vcs_info:68> vcs=cvs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs \n+vcs_info:77> VCS_INFO_detect_cvs\n+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cvs (/*)\n+VCS_INFO_check_com:7> case cvs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cvs:9> return 1\n+vcs_info:68> vcs=darcs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs \n+vcs_info:77> VCS_INFO_detect_darcs\n+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case darcs (/*)\n+VCS_INFO_check_com:7> case darcs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_darcs:9> return 1\n+vcs_info:68> vcs=fossil\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil \n+vcs_info:77> VCS_INFO_detect_fossil\n+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case fossil (/*)\n+VCS_INFO_check_com:7> case fossil (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_fossil:9> return 1\n+vcs_info:68> vcs=git\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git \n+vcs_info:77> VCS_INFO_detect_git\n+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_git:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_git:9> VCS_INFO_check_com git\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case git (/*)\n+VCS_INFO_check_com:7> case git (*)\n+VCS_INFO_check_com:12> (( 1 ))\n+VCS_INFO_check_com:12> return 0\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]='' \n+VCS_INFO_detect_git:15> return 1\n+vcs_info:68> vcs=hg\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg \n+vcs_info:77> VCS_INFO_detect_hg\n+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case hg (/*)\n+VCS_INFO_check_com:7> case hg (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_hg:9> return 1\n+vcs_info:68> vcs=mtn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn \n+vcs_info:77> VCS_INFO_detect_mtn\n+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case mtn (/*)\n+VCS_INFO_check_com:7> case mtn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_mtn:9> return 1\n+vcs_info:68> vcs=p4\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4 \n+vcs_info:77> VCS_INFO_detect_p4\n+VCS_INFO_detect_p4:1> local serverport p4where\n+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server\n+VCS_INFO_detect_p4:23> [[ -n '' ]]\n+VCS_INFO_detect_p4:23> return 1\n+vcs_info:68> vcs=svk\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk \n+vcs_info:77> VCS_INFO_detect_svk\n+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob\n+VCS_INFO_detect_svk:17> local -i fhash\n+VCS_INFO_detect_svk:18> fhash=0 \n+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svk (/*)\n+VCS_INFO_check_com:7> case svk (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svk:20> return 1\n+vcs_info:68> vcs=svn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn \n+vcs_info:77> VCS_INFO_detect_svn\n+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svn (/*)\n+VCS_INFO_check_com:7> case svn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svn:9> return 1\n+vcs_info:68> vcs=tla\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla \n+vcs_info:77> VCS_INFO_detect_tla\n+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case tla (/*)\n+VCS_INFO_check_com:7> case tla (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_tla:9> return 1\n+vcs_info:80> (( found == 0 ))\n+vcs_info:81> vcs=-quilt- \n+vcs_info:81> quiltmode=standalone \n+vcs_info:82> VCS_INFO_quilt standalone\n+VCS_INFO_quilt:3> (( 1 ))\n+VCS_INFO_quilt:24> (( 1 ))\n+VCS_INFO_quilt:63> (( 1 ))\n+VCS_INFO_quilt:86> (( 1 ))\n+VCS_INFO_quilt:92> emulate -L zsh\n+VCS_INFO_quilt:93> setopt extendedglob\n+VCS_INFO_quilt:94> local mode=standalone\n+VCS_INFO_quilt:95> local patches pc qstring root\n+VCS_INFO_quilt:96> local -i ret\n+VCS_INFO_quilt:97> local context\n+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env\n+VCS_INFO_quilt:99> local -A hook_com\n+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all- \n+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt\n+VCS_INFO_quilt:102> return 1\n+vcs_info:82> VCS_INFO_set --nvcs\n+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset\n+VCS_INFO_set:8> local -i i j\n+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]\n+VCS_INFO_set:11> [[ '' == -preinit- ]]\n+VCS_INFO_set:12> i=0\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='\n+VCS_INFO_set:12> i=1\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='\n+VCS_INFO_set:15> VCS_INFO_nvcsformats\n+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit\n+VCS_INFO_nvcsformats:6> local c v rr\n+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]\n+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs\n+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))\n+VCS_INFO_nvcsformats:15> return 0\n+VCS_INFO_set:16> [[ '' != -preinit- ]]\n+VCS_INFO_set:16> VCS_INFO_hook no-vcs\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=no-vcs \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+VCS_INFO_set:19> (( 0 - 1 < 0 ))\n+VCS_INFO_set:19> return 0\n+vcs_info:83> return 0\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ set +x \n+-zsh:67> set +x\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-11\n[2026-05-13 11:40:52] ========================================\n[2026-05-13 11:40:52] Screenpipe sync starting for: 2026-05-11\n[2026-05-13 11:40:52] ========================================\n\n[+00m00s] ▶ Preflight checks\n Source DB: OK (4.5G)\n NAS mount: OK /Volumes/screenpipe\n[2026-05-13 11:40:53] Date 2026-05-11 already has 6857 frames in archive — skipping DB sync\n Data dir: OK (283 files, 318M)\n\n[+00m01s] ▶ Copying data folder for 2026-05-11\n rsync 2026-05-11/ → NAS ✓ 0m01s (283 files, 318M)\n\n[+00m02s] ▶ Copying audio files for 2026-05-11\n rsync audio files → NAS skipped (no audio files for date)\n\n[+00m02s] ▶ Copying screenpipe logs for 2026-05-11\n rsync logs → NAS ✓ 1 file(s), 520K\n\n[2026-05-13 11:40:54] Archive DB size: 1.3G\n[2026-05-13 11:40:54] Total time: 0m2s\n[2026-05-13 11:40:54] Sync complete for 2026-05-11\n[2026-05-13 11:40:54] ========================================\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12\n[2026-05-13 11:41:34] ========================================\n UW PICO 5.09 New Buffer \n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n [ Read 297 lines ] \n^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos \n^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell \n UW PICO 5.09 File: screenpipe_sync_db.sh \n\n Pico Help Text\n \n Pico is designed to be a simple, easy-to-use text editor with a\n layout very similar to the Alpine mailer. The status line at the\n top of the display shows pico's version, the current file being\n edited and whether or not there are outstanding modifications\n that have not been saved. The third line from the bottom is used\n to report informational messages and for additional command input.\n The bottom two lines list the available editing commands.\n \n Each character typed is automatically inserted into the buffer\n at the current cursor position. Editing commands and cursor\n movement (besides arrow keys) are given to pico by typing\n special control-key sequences. A caret, '^', is used to denote\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000the control key, sometimes marked \"CTRL\", so the CTRL-q key\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000combination is written as ^Q.\n \n The following functions are available in pico (where applicable,\n corresponding function key commands are in parentheses).\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^G (F1) Display this help text.\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^F move Forward a character.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^B move Backward a character.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^P move to the Previous line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^N move to the Next line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^A move to the beginning of the current line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^E move to the End of the current line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^V (F8) move forward a page of text.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^Y (F7) move backward a page of text.\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^W (F6) Search for (where is) text, neglecting case.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^L Refresh the display.\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^D Delete the character at the cursor position.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^^ Mark cursor position as beginning of selected text.\n Note: Setting mark when already set unselects text.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^K (F9) Cut selected text (displayed in inverse characters).\n Note: The selected text's boundary on the cursor side\n ends at the left edge of the cursor. So, with \n\n \n^X Exit Help ^V Next Pg \n[2026-05-13 11:41:34] Screenpipe sync starting for: 2026-05-12\n[2026-05-13 11:41:34] ========================================\n\n[+00m00s] ▶ Preflight checks\n Source DB: OK (4.5G)\n NAS mount: OK /Volumes/screenpipe\n Archive DB: exists (1.3G)\n Data dir: OK (220 files, 303M)\n\n[+00m00s] ▶ Counting source rows for 2026-05-12\n frames: 7274\n elements: 853406\n ui_events: 7044\n ocr_text: 2306\n meetings: 3\n audio_chunks: 1113\n audio_transcriptions: 319\n\n[+00m01s] ▶ Initialising tables, indexes, FTS\n creating tables ✓ 0m00s\n creating indexes ✓ 0m00s\n creating FTS tables ✓ 0m00s\n\n[+00m01s] ▶ Syncing vision data for 2026-05-12\n video_chunks ✓ 0m01s\n frames (7274 rows) ✓ 1m05s\n ocr_text (2306 rows) ✓ 0m49s\n ui_events (7044 rows) ✓ 0m01s\n elements (853406 rows) ✓ 0m46s\n meetings (3 rows) ✓ 0m00s\n\n[+02m43s] ▶ Syncing audio data for 2026-05-12\n audio_chunks (1113 rows) ✓ 0m00s\n audio_transcriptions (319 rows) ✓ 0m01s\n\n[+02m44s] ▶ Updating FTS indexes\n elements_fts ✓ 0m50s\n frames_fts ✓ 1m13s\n ui_events_fts ✓ 0m01s\n audio_transcriptions_fts ✓ 0m01s\n\n[+04m49s] ▶ Verifying DB\n/Users/lukas/.screenpipe/scripts/lib/screenpipe_sync_db.sh: line 220: target_DATE: unbound variable\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh \nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12\n[2026-05-13 11:53:26] ========================================\n[2026-05-13 11:53:26] Screenpipe sync starting for: 2026-05-12\n[2026-05-13 11:53:26] ========================================\n\n[+00m00s] ▶ Preflight checks\n Source DB: OK (4.6G)\n NAS mount: OK /Volumes/screenpipe\n Archive DB: will be created\n Data dir: OK (220 files, 303M)\n\n[+00m00s] ▶ Counting source rows for 2026-05-12\n frames: 7274\n elements: 853406\n ui_events: 7044\n ocr_text: 2306\n meetings: 3\n audio_chunks: 1113\n audio_transcriptions: 319\n\n[+00m00s] ▶ Initialising tables, indexes, FTS\n creating tables ✓ 0m00s\n creating indexes ✓ 0m01s\n creating FTS tables ✓ 0m00s\n\n[+00m02s] ▶ Syncing vision data for 2026-05-12\n video_chunks ✓ 0m00s\n frames (7274 rows) ✓ 1m03s\n ocr_text (2306 rows) ✓ 0m48s\n ui_events (7044 rows) ✓ 0m01s\n elements (853406 rows) ✓ 0m46s\n meetings (3 rows) ✓ 0m00s\n\n[+02m40s] ▶ Syncing audio data for 2026-05-12\n audio_chunks (1113 rows) ✓ 0m00s\n audio_transcriptions (319 rows) ✓ 0m00s\n\n[+02m40s] ▶ Updating FTS indexes\n elements_fts ✓ 0m32s\n frames_fts ✓ 1m08s\n ui_events_fts ✓ 0m00s\n audio_transcriptions_fts ✓ 0m00s\n\n[+04m20s] ▶ Verifying DB\n frames: 7274 / 7274 ✓\n elements: 853406 / 853406 ✓\n ui_events: 7044 / 7044 ✓\n ocr_text: 2306 / 2306 ✓\n meetings: 3 / 3 ✓\n audio_chunks: 1113 / 1113 ✓\n audio_transcriptions: 319 / 319 ✓\n\n[+04m29s] ▶ Copying data folder for 2026-05-12\n rsync 2026-05-12/ → NAS ✓ 0m20s (220 files, 295M)\n\n[+04m49s] ▶ Copying audio files for 2026-05-12\n rsync audio files → NAS skipped (no audio files for date)\n\n[+04m49s] ▶ Copying screenpipe logs for 2026-05-12\n rsync logs → NAS ✓ 1 file(s), 304K\n\n[2026-05-13 11:58:15] Archive DB size: 763M\n[2026-05-13 11:58:15] Total time: 4m49s\n[2026-05-13 11:58:15] Sync complete for 2026-05-12\n[2026-05-13 11:58:15] ========================================\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $","depth":4,"on_screen":true,"value":"+VCS_INFO_quilt:24> (( 1 ))\n+VCS_INFO_quilt:63> (( 1 ))\n+VCS_INFO_quilt:86> (( 1 ))\n+VCS_INFO_quilt:92> emulate -L zsh\n+VCS_INFO_quilt:93> setopt extendedglob\n+VCS_INFO_quilt:94> local mode=standalone\n+VCS_INFO_quilt:95> local patches pc qstring root\n+VCS_INFO_quilt:96> local -i ret\n+VCS_INFO_quilt:97> local context\n+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env\n+VCS_INFO_quilt:99> local -A hook_com\n+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all- \n+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt\n+VCS_INFO_quilt:102> return 1\n+vcs_info:82> VCS_INFO_set --nvcs\n+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset\n+VCS_INFO_set:8> local -i i j\n+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]\n+VCS_INFO_set:11> [[ '' == -preinit- ]]\n+VCS_INFO_set:12> i=0\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='\n+VCS_INFO_set:12> i=1\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='\n+VCS_INFO_set:15> VCS_INFO_nvcsformats\n+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit\n+VCS_INFO_nvcsformats:6> local c v rr\n+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]\n+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs\n+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))\n+VCS_INFO_nvcsformats:15> return 0\n+VCS_INFO_set:16> [[ '' != -preinit- ]]\n+VCS_INFO_set:16> VCS_INFO_hook no-vcs\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=no-vcs \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+VCS_INFO_set:19> (( 0 - 1 < 0 ))\n+VCS_INFO_set:19> return 0\n+vcs_info:83> return 0\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts $ cd lib \n+-zsh:65> cd lib\n+precmd:0> vcs_info \n+vcs_info:1> emulate -L zsh\n+vcs_info:2> setopt extendedglob NO_warn_create_global\n+vcs_info:4> [[ -r . ]]\n+vcs_info:6> local pat\n+vcs_info:7> local -i found retval\n+vcs_info:8> local -a enabled disabled dps\n+vcs_info:9> local usercontext vcs rrn quiltmode\n+vcs_info:10> local -x LC_MESSAGES\n+vcs_info:11> local -i maxexports\n+vcs_info:12> local -a msgs\n+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data\n+vcs_info:21> LC_MESSAGES=C \n+vcs_info:22> [[ -n '' ]]\n+vcs_info:27> vcs=-init- \n+vcs_info:27> rrn=-all- \n+vcs_info:27> quiltmode=addon \n+vcs_info:28> usercontext=default \n+vcs_info:30> VCS_INFO_hook start-up\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=start-up \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+vcs_info:31> retval=0 \n+vcs_info:32> (( retval == 1 ))\n+vcs_info:34> (( retval == 2 ))\n+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled\n+vcs_info:43> (( 0 == 0 ))\n+vcs_info:43> enabled=( all ) \n+vcs_info:45> [[ -n '' ]]\n+vcs_info:50> [[ -n all ]]\n+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla ) \n+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled\n+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps\n+vcs_info:65> VCS_INFO_maxexports\n+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset\n+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports\n+VCS_INFO_maxexports:7> maxexports=2 \n+VCS_INFO_maxexports:8> [[ 2 != <-> ]]\n+VCS_INFO_maxexports:8> (( maxexports < 1 ))\n+VCS_INFO_maxexports:13> return 0\n+vcs_info:67> (( found = 0 ))\n+vcs_info:68> vcs=bzr\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr \n+vcs_info:77> VCS_INFO_detect_bzr\n+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case bzr (/*)\n+VCS_INFO_check_com:7> case bzr (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_bzr:9> return 1\n+vcs_info:68> vcs=cdv\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv \n+vcs_info:77> VCS_INFO_detect_cdv\n+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cdv (/*)\n+VCS_INFO_check_com:7> case cdv (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cdv:9> return 1\n+vcs_info:68> vcs=cvs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs \n+vcs_info:77> VCS_INFO_detect_cvs\n+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cvs (/*)\n+VCS_INFO_check_com:7> case cvs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cvs:9> return 1\n+vcs_info:68> vcs=darcs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs \n+vcs_info:77> VCS_INFO_detect_darcs\n+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case darcs (/*)\n+VCS_INFO_check_com:7> case darcs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_darcs:9> return 1\n+vcs_info:68> vcs=fossil\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil \n+vcs_info:77> VCS_INFO_detect_fossil\n+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case fossil (/*)\n+VCS_INFO_check_com:7> case fossil (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_fossil:9> return 1\n+vcs_info:68> vcs=git\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git \n+vcs_info:77> VCS_INFO_detect_git\n+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_git:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_git:9> VCS_INFO_check_com git\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case git (/*)\n+VCS_INFO_check_com:7> case git (*)\n+VCS_INFO_check_com:12> (( 1 ))\n+VCS_INFO_check_com:12> return 0\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]='' \n+VCS_INFO_detect_git:15> return 1\n+vcs_info:68> vcs=hg\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg \n+vcs_info:77> VCS_INFO_detect_hg\n+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case hg (/*)\n+VCS_INFO_check_com:7> case hg (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_hg:9> return 1\n+vcs_info:68> vcs=mtn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn \n+vcs_info:77> VCS_INFO_detect_mtn\n+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case mtn (/*)\n+VCS_INFO_check_com:7> case mtn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_mtn:9> return 1\n+vcs_info:68> vcs=p4\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4 \n+vcs_info:77> VCS_INFO_detect_p4\n+VCS_INFO_detect_p4:1> local serverport p4where\n+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server\n+VCS_INFO_detect_p4:23> [[ -n '' ]]\n+VCS_INFO_detect_p4:23> return 1\n+vcs_info:68> vcs=svk\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk \n+vcs_info:77> VCS_INFO_detect_svk\n+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob\n+VCS_INFO_detect_svk:17> local -i fhash\n+VCS_INFO_detect_svk:18> fhash=0 \n+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svk (/*)\n+VCS_INFO_check_com:7> case svk (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svk:20> return 1\n+vcs_info:68> vcs=svn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn \n+vcs_info:77> VCS_INFO_detect_svn\n+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svn (/*)\n+VCS_INFO_check_com:7> case svn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svn:9> return 1\n+vcs_info:68> vcs=tla\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla \n+vcs_info:77> VCS_INFO_detect_tla\n+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case tla (/*)\n+VCS_INFO_check_com:7> case tla (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_tla:9> return 1\n+vcs_info:80> (( found == 0 ))\n+vcs_info:81> vcs=-quilt- \n+vcs_info:81> quiltmode=standalone \n+vcs_info:82> VCS_INFO_quilt standalone\n+VCS_INFO_quilt:3> (( 1 ))\n+VCS_INFO_quilt:24> (( 1 ))\n+VCS_INFO_quilt:63> (( 1 ))\n+VCS_INFO_quilt:86> (( 1 ))\n+VCS_INFO_quilt:92> emulate -L zsh\n+VCS_INFO_quilt:93> setopt extendedglob\n+VCS_INFO_quilt:94> local mode=standalone\n+VCS_INFO_quilt:95> local patches pc qstring root\n+VCS_INFO_quilt:96> local -i ret\n+VCS_INFO_quilt:97> local context\n+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env\n+VCS_INFO_quilt:99> local -A hook_com\n+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all- \n+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt\n UW PICO 5.09 New Buffer \n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n [ Read 297 lines ] \n^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos \n^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell \n+VCS_INFO_quilt:102> return 1\n+vcs_info:82> VCS_INFO_set --nvcs\n+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset\n+VCS_INFO_set:8> local -i i j\n+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]\n+VCS_INFO_set:11> [[ '' == -preinit- ]]\n+VCS_INFO_set:12> i=0\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='\n+VCS_INFO_set:12> i=1\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='\n+VCS_INFO_set:15> VCS_INFO_nvcsformats\n+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit\n+VCS_INFO_nvcsformats:6> local c v rr\n+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]\n+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs\n+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))\n+VCS_INFO_nvcsformats:15> return 0\n+VCS_INFO_set:16> [[ '' != -preinit- ]]\n+VCS_INFO_set:16> VCS_INFO_hook no-vcs\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=no-vcs \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+VCS_INFO_set:19> (( 0 - 1 < 0 ))\n+VCS_INFO_set:19> return 0\n+vcs_info:83> return 0\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh\n+-zsh:66> nano screenpipe_sync_db.sh\n+precmd:0> vcs_info \n+vcs_info:1> emulate -L zsh\n+vcs_info:2> setopt extendedglob NO_warn_create_global\n+vcs_info:4> [[ -r . ]]\n+vcs_info:6> local pat\n+vcs_info:7> local -i found retval\n+vcs_info:8> local -a enabled disabled dps\n+vcs_info:9> local usercontext vcs rrn quiltmode\n+vcs_info:10> local -x LC_MESSAGES\n+vcs_info:11> local -i maxexports\n+vcs_info:12> local -a msgs\n+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data\n+vcs_info:21> LC_MESSAGES=C \n+vcs_info:22> [[ -n '' ]]\n+vcs_info:27> vcs=-init- \n+vcs_info:27> rrn=-all- \n+vcs_info:27> quiltmode=addon \n+vcs_info:28> usercontext=default \n+vcs_info:30> VCS_INFO_hook start-up\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=start-up \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+vcs_info:31> retval=0 \n+vcs_info:32> (( retval == 1 ))\n+vcs_info:34> (( retval == 2 ))\n+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled\n+vcs_info:43> (( 0 == 0 ))\n+vcs_info:43> enabled=( all ) \n+vcs_info:45> [[ -n '' ]]\n+vcs_info:50> [[ -n all ]]\n+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla ) \n+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled\n+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps\n+vcs_info:65> VCS_INFO_maxexports\n+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset\n+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports\n+VCS_INFO_maxexports:7> maxexports=2 \n+VCS_INFO_maxexports:8> [[ 2 != <-> ]]\n+VCS_INFO_maxexports:8> (( maxexports < 1 ))\n+VCS_INFO_maxexports:13> return 0\n+vcs_info:67> (( found = 0 ))\n+vcs_info:68> vcs=bzr\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr \n+vcs_info:77> VCS_INFO_detect_bzr\n+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case bzr (/*)\n+VCS_INFO_check_com:7> case bzr (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_bzr:9> return 1\n+vcs_info:68> vcs=cdv\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv \n+vcs_info:77> VCS_INFO_detect_cdv\n+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cdv (/*)\n+VCS_INFO_check_com:7> case cdv (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cdv:9> return 1\n+vcs_info:68> vcs=cvs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs \n+vcs_info:77> VCS_INFO_detect_cvs\n+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case cvs (/*)\n+VCS_INFO_check_com:7> case cvs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_cvs:9> return 1\n+vcs_info:68> vcs=darcs\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs \n+vcs_info:77> VCS_INFO_detect_darcs\n+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case darcs (/*)\n+VCS_INFO_check_com:7> case darcs (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_darcs:9> return 1\n+vcs_info:68> vcs=fossil\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil \n+vcs_info:77> VCS_INFO_detect_fossil\n+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case fossil (/*)\n+VCS_INFO_check_com:7> case fossil (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_fossil:9> return 1\n+vcs_info:68> vcs=git\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git \n+vcs_info:77> VCS_INFO_detect_git\n+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_git:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_git:9> VCS_INFO_check_com git\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case git (/*)\n+VCS_INFO_check_com:7> case git (*)\n+VCS_INFO_check_com:12> (( 1 ))\n+VCS_INFO_check_com:12> return 0\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir\n+VCS_INFO_detect_git:9> vcs_comm[gitdir]='' \n+VCS_INFO_detect_git:15> return 1\n+vcs_info:68> vcs=hg\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg \n+vcs_info:77> VCS_INFO_detect_hg\n+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case hg (/*)\n+VCS_INFO_check_com:7> case hg (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_hg:9> return 1\n+vcs_info:68> vcs=mtn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn \n+vcs_info:77> VCS_INFO_detect_mtn\n+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case mtn (/*)\n+VCS_INFO_check_com:7> case mtn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_mtn:9> return 1\n+vcs_info:68> vcs=p4\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4 \n+vcs_info:77> VCS_INFO_detect_p4\n+VCS_INFO_detect_p4:1> local serverport p4where\n+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server\n+VCS_INFO_detect_p4:23> [[ -n '' ]]\n+VCS_INFO_detect_p4:23> return 1\n+vcs_info:68> vcs=svk\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk \n+vcs_info:77> VCS_INFO_detect_svk\n+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob\n+VCS_INFO_detect_svk:17> local -i fhash\n+VCS_INFO_detect_svk:18> fhash=0 \n+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svk (/*)\n+VCS_INFO_check_com:7> case svk (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svk:20> return 1\n+vcs_info:68> vcs=svn\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn \n+vcs_info:77> VCS_INFO_detect_svn\n+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case svn (/*)\n+VCS_INFO_check_com:7> case svn (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_svn:9> return 1\n+vcs_info:68> vcs=tla\n+vcs_info:69> [[ -n '' ]]\n+vcs_info:70> (( 1 == 0 ))\n+vcs_info:75> vcs_comm=( ) \n+vcs_info:76> VCS_INFO_get_cmd\n+VCS_INFO_get_cmd:4> local cmd\n+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd\n+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla \n+vcs_info:77> VCS_INFO_detect_tla\n+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]\n+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla\n+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit\n+VCS_INFO_check_com:7> case tla (/*)\n+VCS_INFO_check_com:7> case tla (*)\n+VCS_INFO_check_com:12> (( 0 ))\n+VCS_INFO_check_com:15> return 1\n+VCS_INFO_detect_tla:9> return 1\n+vcs_info:80> (( found == 0 ))\n+vcs_info:81> vcs=-quilt- \n+vcs_info:81> quiltmode=standalone \n+vcs_info:82> VCS_INFO_quilt standalone\n+VCS_INFO_quilt:3> (( 1 ))\n+VCS_INFO_quilt:24> (( 1 ))\n+VCS_INFO_quilt:63> (( 1 ))\n+VCS_INFO_quilt:86> (( 1 ))\n+VCS_INFO_quilt:92> emulate -L zsh\n+VCS_INFO_quilt:93> setopt extendedglob\n+VCS_INFO_quilt:94> local mode=standalone\n+VCS_INFO_quilt:95> local patches pc qstring root\n+VCS_INFO_quilt:96> local -i ret\n+VCS_INFO_quilt:97> local context\n+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env\n+VCS_INFO_quilt:99> local -A hook_com\n+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all- \n+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt\n+VCS_INFO_quilt:102> return 1\n+vcs_info:82> VCS_INFO_set --nvcs\n+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset\n+VCS_INFO_set:8> local -i i j\n+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]\n+VCS_INFO_set:11> [[ '' == -preinit- ]]\n+VCS_INFO_set:12> i=0\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='\n+VCS_INFO_set:12> i=1\n+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='\n+VCS_INFO_set:15> VCS_INFO_nvcsformats\n+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit\n+VCS_INFO_nvcsformats:6> local c v rr\n+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]\n+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs\n+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))\n+VCS_INFO_nvcsformats:15> return 0\n+VCS_INFO_set:16> [[ '' != -preinit- ]]\n+VCS_INFO_set:16> VCS_INFO_hook no-vcs\n+VCS_INFO_hook:5> local hook static func\n+VCS_INFO_hook:6> local context hook_name\n+VCS_INFO_hook:7> local -i ret\n+VCS_INFO_hook:8> local -a hooks tmp\n+VCS_INFO_hook:9> local -i debug\n+VCS_INFO_hook:11> ret=0 \n+VCS_INFO_hook:12> hook_name=no-vcs \n+VCS_INFO_hook:13> shift\n+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all- \n+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs \n+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug\n+VCS_INFO_hook:17> debug=0 \n+VCS_INFO_hook:18> (( debug ))\n+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks\n+VCS_INFO_hook:25> (( debug ))\n+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp\n+VCS_INFO_hook:29> (( debug ))\n+VCS_INFO_hook:32> hooks+=( ) \n+VCS_INFO_hook:33> (( 0 == 0 ))\n+VCS_INFO_hook:33> return 0\n+VCS_INFO_set:19> (( 0 - 1 < 0 ))\n+VCS_INFO_set:19> return 0\n+vcs_info:83> return 0\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ set +x \n+-zsh:67> set +x\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-11\n[2026-05-13 11:40:52] ========================================\n[2026-05-13 11:40:52] Screenpipe sync starting for: 2026-05-11\n[2026-05-13 11:40:52] ========================================\n\n[+00m00s] ▶ Preflight checks\n Source DB: OK (4.5G)\n NAS mount: OK /Volumes/screenpipe\n[2026-05-13 11:40:53] Date 2026-05-11 already has 6857 frames in archive — skipping DB sync\n Data dir: OK (283 files, 318M)\n\n[+00m01s] ▶ Copying data folder for 2026-05-11\n rsync 2026-05-11/ → NAS ✓ 0m01s (283 files, 318M)\n\n[+00m02s] ▶ Copying audio files for 2026-05-11\n rsync audio files → NAS skipped (no audio files for date)\n\n[+00m02s] ▶ Copying screenpipe logs for 2026-05-11\n rsync logs → NAS ✓ 1 file(s), 520K\n\n[2026-05-13 11:40:54] Archive DB size: 1.3G\n[2026-05-13 11:40:54] Total time: 0m2s\n[2026-05-13 11:40:54] Sync complete for 2026-05-11\n[2026-05-13 11:40:54] ========================================\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12\n[2026-05-13 11:41:34] ========================================\n UW PICO 5.09 New Buffer \n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n [ Read 297 lines ] \n^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos \n^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell \n UW PICO 5.09 File: screenpipe_sync_db.sh \n\n Pico Help Text\n \n Pico is designed to be a simple, easy-to-use text editor with a\n layout very similar to the Alpine mailer. The status line at the\n top of the display shows pico's version, the current file being\n edited and whether or not there are outstanding modifications\n that have not been saved. The third line from the bottom is used\n to report informational messages and for additional command input.\n The bottom two lines list the available editing commands.\n \n Each character typed is automatically inserted into the buffer\n at the current cursor position. Editing commands and cursor\n movement (besides arrow keys) are given to pico by typing\n special control-key sequences. A caret, '^', is used to denote\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000the control key, sometimes marked \"CTRL\", so the CTRL-q key\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000combination is written as ^Q.\n \n The following functions are available in pico (where applicable,\n corresponding function key commands are in parentheses).\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^G (F1) Display this help text.\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^F move Forward a character.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^B move Backward a character.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^P move to the Previous line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^N move to the Next line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^A move to the beginning of the current line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^E move to the End of the current line.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^V (F8) move forward a page of text.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^Y (F7) move backward a page of text.\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^W (F6) Search for (where is) text, neglecting case.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^L Refresh the display.\n \n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^D Delete the character at the cursor position.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^^ Mark cursor position as beginning of selected text.\n Note: Setting mark when already set unselects text.\n\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000^K (F9) Cut selected text (displayed in inverse characters).\n Note: The selected text's boundary on the cursor side\n ends at the left edge of the cursor. So, with \n\n \n^X Exit Help ^V Next Pg \n[2026-05-13 11:41:34] Screenpipe sync starting for: 2026-05-12\n[2026-05-13 11:41:34] ========================================\n\n[+00m00s] ▶ Preflight checks\n Source DB: OK (4.5G)\n NAS mount: OK /Volumes/screenpipe\n Archive DB: exists (1.3G)\n Data dir: OK (220 files, 303M)\n\n[+00m00s] ▶ Counting source rows for 2026-05-12\n frames: 7274\n elements: 853406\n ui_events: 7044\n ocr_text: 2306\n meetings: 3\n audio_chunks: 1113\n audio_transcriptions: 319\n\n[+00m01s] ▶ Initialising tables, indexes, FTS\n creating tables ✓ 0m00s\n creating indexes ✓ 0m00s\n creating FTS tables ✓ 0m00s\n\n[+00m01s] ▶ Syncing vision data for 2026-05-12\n video_chunks ✓ 0m01s\n frames (7274 rows) ✓ 1m05s\n ocr_text (2306 rows) ✓ 0m49s\n ui_events (7044 rows) ✓ 0m01s\n elements (853406 rows) ✓ 0m46s\n meetings (3 rows) ✓ 0m00s\n\n[+02m43s] ▶ Syncing audio data for 2026-05-12\n audio_chunks (1113 rows) ✓ 0m00s\n audio_transcriptions (319 rows) ✓ 0m01s\n\n[+02m44s] ▶ Updating FTS indexes\n elements_fts ✓ 0m50s\n frames_fts ✓ 1m13s\n ui_events_fts ✓ 0m01s\n audio_transcriptions_fts ✓ 0m01s\n\n[+04m49s] ▶ Verifying DB\n/Users/lukas/.screenpipe/scripts/lib/screenpipe_sync_db.sh: line 220: target_DATE: unbound variable\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh \nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12\n[2026-05-13 11:53:26] ========================================\n[2026-05-13 11:53:26] Screenpipe sync starting for: 2026-05-12\n[2026-05-13 11:53:26] ========================================\n\n[+00m00s] ▶ Preflight checks\n Source DB: OK (4.6G)\n NAS mount: OK /Volumes/screenpipe\n Archive DB: will be created\n Data dir: OK (220 files, 303M)\n\n[+00m00s] ▶ Counting source rows for 2026-05-12\n frames: 7274\n elements: 853406\n ui_events: 7044\n ocr_text: 2306\n meetings: 3\n audio_chunks: 1113\n audio_transcriptions: 319\n\n[+00m00s] ▶ Initialising tables, indexes, FTS\n creating tables ✓ 0m00s\n creating indexes ✓ 0m01s\n creating FTS tables ✓ 0m00s\n\n[+00m02s] ▶ Syncing vision data for 2026-05-12\n video_chunks ✓ 0m00s\n frames (7274 rows) ✓ 1m03s\n ocr_text (2306 rows) ✓ 0m48s\n ui_events (7044 rows) ✓ 0m01s\n elements (853406 rows) ✓ 0m46s\n meetings (3 rows) ✓ 0m00s\n\n[+02m40s] ▶ Syncing audio data for 2026-05-12\n audio_chunks (1113 rows) ✓ 0m00s\n audio_transcriptions (319 rows) ✓ 0m00s\n\n[+02m40s] ▶ Updating FTS indexes\n elements_fts ✓ 0m32s\n frames_fts ✓ 1m08s\n ui_events_fts ✓ 0m00s\n audio_transcriptions_fts ✓ 0m00s\n\n[+04m20s] ▶ Verifying DB\n frames: 7274 / 7274 ✓\n elements: 853406 / 853406 ✓\n ui_events: 7044 / 7044 ✓\n ocr_text: 2306 / 2306 ✓\n meetings: 3 / 3 ✓\n audio_chunks: 1113 / 1113 ✓\n audio_transcriptions: 319 / 319 ✓\n\n[+04m29s] ▶ Copying data folder for 2026-05-12\n rsync 2026-05-12/ → NAS ✓ 0m20s (220 files, 295M)\n\n[+04m49s] ▶ Copying audio files for 2026-05-12\n rsync audio files → NAS skipped (no audio files for date)\n\n[+04m49s] ▶ Copying screenpipe logs for 2026-05-12\n rsync logs → NAS ✓ 1 file(s), 304K\n\n[2026-05-13 11:58:15] Archive DB size: 763M\n[2026-05-13 11:58:15] Total time: 4m49s\n[2026-05-13 11:58:15] Sync complete for 2026-05-12\n[2026-05-13 11:58:15] ========================================\nlukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $","is_focused":true},{"role":"AXRadioButton","text":"DOCKER","depth":2,"bounds":{"left":0.0,"top":0.05888889,"width":0.19722222,"height":0.026666667},"on_screen":true,"role_description":"radio button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close Tab","depth":3,"bounds":{"left":0.004166667,"top":0.06333333,"width":0.011111111,"height":0.017777778},"on_screen":true,"role_description":"button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXRadioButton","text":"DEV (-zsh)","depth":2,"bounds":{"left":0.19722222,"top":0.05888889,"width":0.196875,"height":0.026666667},"on_screen":true,"role_description":"radio button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close Tab","depth":3,"bounds":{"left":0.2013889,"top":0.06333333,"width":0.011111111,"height":0.017777778},"on_screen":true,"role_description":"button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXRadioButton","text":"APP (-zsh)","depth":2,"bounds":{"left":0.3940972,"top":0.05888889,"width":0.196875,"height":0.026666667},"on_screen":true,"role_description":"radio button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close Tab","depth":3,"bounds":{"left":0.3982639,"top":0.06333333,"width":0.011111111,"height":0.017777778},"on_screen":true,"role_description":"button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXRadioButton","text":"-zsh","depth":2,"bounds":{"left":0.59097224,"top":0.05888889,"width":0.196875,"height":0.026666667},"on_screen":true,"role_description":"radio button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close Tab","depth":3,"bounds":{"left":0.5951389,"top":0.06333333,"width":0.011111111,"height":0.017777778},"on_screen":true,"role_description":"button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXRadioButton","text":"screenpipe\"","depth":2,"bounds":{"left":0.7878472,"top":0.05888889,"width":0.196875,"height":0.026666667},"on_screen":true,"role_description":"radio button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close Tab","depth":3,"bounds":{"left":0.7920139,"top":0.06333333,"width":0.011111111,"height":0.017777778},"on_screen":true,"role_description":"button","is_enabled":false,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"⌥⌘1","depth":1,"bounds":{"left":0.95625,"top":0.032222223,"width":0.03888889,"height":0.018888889},"on_screen":true,"automation_id":"_NS:8","role_description":"text"},{"role":"AXStaticText","text":"-zsh","depth":1,"bounds":{"left":0.4888889,"top":0.033333335,"width":0.022916667,"height":0.017777778},"on_screen":true,"role_description":"text"}]...
|
4223095973093514271
|
1485174741784092104
|
app_switch
|
accessibility
|
NULL
|
+VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:6 +VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:63> (( 1 ))
+VCS_INFO_quilt:86> (( 1 ))
+VCS_INFO_quilt:92> emulate -L zsh
+VCS_INFO_quilt:93> setopt extendedglob
+VCS_INFO_quilt:94> local mode=standalone
+VCS_INFO_quilt:95> local patches pc qstring root
+VCS_INFO_quilt:96> local -i ret
+VCS_INFO_quilt:97> local context
+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env
+VCS_INFO_quilt:99> local -A hook_com
+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all-
+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt
+VCS_INFO_quilt:102> return 1
+vcs_info:82> VCS_INFO_set --nvcs
+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset
+VCS_INFO_set:8> local -i i j
+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]
+VCS_INFO_set:11> [[ '' == -preinit- ]]
+VCS_INFO_set:12> i=0
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='
+VCS_INFO_set:12> i=1
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='
+VCS_INFO_set:15> VCS_INFO_nvcsformats
+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit
+VCS_INFO_nvcsformats:6> local c v rr
+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]
+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs
+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))
+VCS_INFO_nvcsformats:15> return 0
+VCS_INFO_set:16> [[ '' != -preinit- ]]
+VCS_INFO_set:16> VCS_INFO_hook no-vcs
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=no-vcs
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs
+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+VCS_INFO_set:19> (( 0 - 1 < 0 ))
+VCS_INFO_set:19> return 0
+vcs_info:83> return 0
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts $ cd lib
+-zsh:65> cd lib
+precmd:0> vcs_info
+vcs_info:1> emulate -L zsh
+vcs_info:2> setopt extendedglob NO_warn_create_global
+vcs_info:4> [[ -r . ]]
+vcs_info:6> local pat
+vcs_info:7> local -i found retval
+vcs_info:8> local -a enabled disabled dps
+vcs_info:9> local usercontext vcs rrn quiltmode
+vcs_info:10> local -x LC_MESSAGES
+vcs_info:11> local -i maxexports
+vcs_info:12> local -a msgs
+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data
+vcs_info:21> LC_MESSAGES=C
+vcs_info:22> [[ -n '' ]]
+vcs_info:27> vcs=-init-
+vcs_info:27> rrn=-all-
+vcs_info:27> quiltmode=addon
+vcs_info:28> usercontext=default
+vcs_info:30> VCS_INFO_hook start-up
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=start-up
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up
+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+vcs_info:31> retval=0
+vcs_info:32> (( retval == 1 ))
+vcs_info:34> (( retval == 2 ))
+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled
+vcs_info:43> (( 0 == 0 ))
+vcs_info:43> enabled=( all )
+vcs_info:45> [[ -n '' ]]
+vcs_info:50> [[ -n all ]]
+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla )
+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled
+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps
+vcs_info:65> VCS_INFO_maxexports
+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset
+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports
+VCS_INFO_maxexports:7> maxexports=2
+VCS_INFO_maxexports:8> [[ 2 != <-> ]]
+VCS_INFO_maxexports:8> (( maxexports < 1 ))
+VCS_INFO_maxexports:13> return 0
+vcs_info:67> (( found = 0 ))
+vcs_info:68> vcs=bzr
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr
+vcs_info:77> VCS_INFO_detect_bzr
+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]
+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case bzr (/*)
+VCS_INFO_check_com:7> case bzr (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_bzr:9> return 1
+vcs_info:68> vcs=cdv
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv
+vcs_info:77> VCS_INFO_detect_cdv
+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cdv (/*)
+VCS_INFO_check_com:7> case cdv (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cdv:9> return 1
+vcs_info:68> vcs=cvs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs
+vcs_info:77> VCS_INFO_detect_cvs
+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cvs (/*)
+VCS_INFO_check_com:7> case cvs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cvs:9> return 1
+vcs_info:68> vcs=darcs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs
+vcs_info:77> VCS_INFO_detect_darcs
+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case darcs (/*)
+VCS_INFO_check_com:7> case darcs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_darcs:9> return 1
+vcs_info:68> vcs=fossil
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil
+vcs_info:77> VCS_INFO_detect_fossil
+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]
+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case fossil (/*)
+VCS_INFO_check_com:7> case fossil (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_fossil:9> return 1
+vcs_info:68> vcs=git
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git
+vcs_info:77> VCS_INFO_detect_git
+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_git:7> [[ '' == --flavours ]]
+VCS_INFO_detect_git:9> VCS_INFO_check_com git
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case git (/*)
+VCS_INFO_check_com:7> case git (*)
+VCS_INFO_check_com:12> (( 1 ))
+VCS_INFO_check_com:12> return 0
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=''
+VCS_INFO_detect_git:15> return 1
+vcs_info:68> vcs=hg
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg
+vcs_info:77> VCS_INFO_detect_hg
+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]
+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case hg (/*)
+VCS_INFO_check_com:7> case hg (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_hg:9> return 1
+vcs_info:68> vcs=mtn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn
+vcs_info:77> VCS_INFO_detect_mtn
+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case mtn (/*)
+VCS_INFO_check_com:7> case mtn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_mtn:9> return 1
+vcs_info:68> vcs=p4
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4
+vcs_info:77> VCS_INFO_detect_p4
+VCS_INFO_detect_p4:1> local serverport p4where
+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server
+VCS_INFO_detect_p4:23> [[ -n '' ]]
+VCS_INFO_detect_p4:23> return 1
+vcs_info:68> vcs=svk
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk
+vcs_info:77> VCS_INFO_detect_svk
+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob
+VCS_INFO_detect_svk:17> local -i fhash
+VCS_INFO_detect_svk:18> fhash=0
+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svk (/*)
+VCS_INFO_check_com:7> case svk (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svk:20> return 1
+vcs_info:68> vcs=svn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn
+vcs_info:77> VCS_INFO_detect_svn
+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svn (/*)
+VCS_INFO_check_com:7> case svn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svn:9> return 1
+vcs_info:68> vcs=tla
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla
+vcs_info:77> VCS_INFO_detect_tla
+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]
+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case tla (/*)
+VCS_INFO_check_com:7> case tla (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_tla:9> return 1
+vcs_info:80> (( found == 0 ))
+vcs_info:81> vcs=-quilt-
+vcs_info:81> quiltmode=standalone
+vcs_info:82> VCS_INFO_quilt standalone
+VCS_INFO_quilt:3> (( 1 ))
+VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:63> (( 1 ))
+VCS_INFO_quilt:86> (( 1 ))
+VCS_INFO_quilt:92> emulate -L zsh
+VCS_INFO_quilt:93> setopt extendedglob
+VCS_INFO_quilt:94> local mode=standalone
+VCS_INFO_quilt:95> local patches pc qstring root
+VCS_INFO_quilt:96> local -i ret
+VCS_INFO_quilt:97> local context
+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env
+VCS_INFO_quilt:99> local -A hook_com
+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all-
+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt
UW PICO 5.09 New Buffer
[ Read 297 lines ]
^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos
^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell
+VCS_INFO_quilt:102> return 1
+vcs_info:82> VCS_INFO_set --nvcs
+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset
+VCS_INFO_set:8> local -i i j
+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]
+VCS_INFO_set:11> [[ '' == -preinit- ]]
+VCS_INFO_set:12> i=0
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='
+VCS_INFO_set:12> i=1
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='
+VCS_INFO_set:15> VCS_INFO_nvcsformats
+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit
+VCS_INFO_nvcsformats:6> local c v rr
+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]
+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs
+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))
+VCS_INFO_nvcsformats:15> return 0
+VCS_INFO_set:16> [[ '' != -preinit- ]]
+VCS_INFO_set:16> VCS_INFO_hook no-vcs
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=no-vcs
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs
+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+VCS_INFO_set:19> (( 0 - 1 < 0 ))
+VCS_INFO_set:19> return 0
+vcs_info:83> return 0
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh
+-zsh:66> nano screenpipe_sync_db.sh
+precmd:0> vcs_info
+vcs_info:1> emulate -L zsh
+vcs_info:2> setopt extendedglob NO_warn_create_global
+vcs_info:4> [[ -r . ]]
+vcs_info:6> local pat
+vcs_info:7> local -i found retval
+vcs_info:8> local -a enabled disabled dps
+vcs_info:9> local usercontext vcs rrn quiltmode
+vcs_info:10> local -x LC_MESSAGES
+vcs_info:11> local -i maxexports
+vcs_info:12> local -a msgs
+vcs_info:19> local -A vcs_comm hook_com backend_misc user_data
+vcs_info:21> LC_MESSAGES=C
+vcs_info:22> [[ -n '' ]]
+vcs_info:27> vcs=-init-
+vcs_info:27> rrn=-all-
+vcs_info:27> quiltmode=addon
+vcs_info:28> usercontext=default
+vcs_info:30> VCS_INFO_hook start-up
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=start-up
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-init-+start-up:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:start-up
+VCS_INFO_hook:17> zstyle -t :vcs_info:-init-+start-up:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:start-up hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-init-+start-up:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+vcs_info:31> retval=0
+vcs_info:32> (( retval == 1 ))
+vcs_info:34> (( retval == 2 ))
+vcs_info:42> zstyle -a :vcs_info:-init-:default:-all- enable enabled
+vcs_info:43> (( 0 == 0 ))
+vcs_info:43> enabled=( all )
+vcs_info:45> [[ -n '' ]]
+vcs_info:50> [[ -n all ]]
+vcs_info:51> enabled=( bzr cdv cvs darcs fossil git hg mtn p4 svk svn tla )
+vcs_info:52> zstyle -a :vcs_info:-init-:default:-all- disable disabled
+vcs_info:55> zstyle -a :vcs_info:-init-:default:-all- disable-patterns dps
+vcs_info:65> VCS_INFO_maxexports
+VCS_INFO_maxexports:5> setopt localoptions NO_shwordsplit unset
+VCS_INFO_maxexports:7> zstyle -s :vcs_info:-init-:default:-all- max-exports maxexports
+VCS_INFO_maxexports:7> maxexports=2
+VCS_INFO_maxexports:8> [[ 2 != <-> ]]
+VCS_INFO_maxexports:8> (( maxexports < 1 ))
+VCS_INFO_maxexports:13> return 0
+vcs_info:67> (( found = 0 ))
+vcs_info:68> vcs=bzr
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:bzr:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=bzr
+vcs_info:77> VCS_INFO_detect_bzr
+VCS_INFO_detect_bzr:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_bzr:7> [[ '' == --flavours ]]
+VCS_INFO_detect_bzr:9> VCS_INFO_check_com bzr
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case bzr (/*)
+VCS_INFO_check_com:7> case bzr (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_bzr:9> return 1
+vcs_info:68> vcs=cdv
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cdv:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cdv
+vcs_info:77> VCS_INFO_detect_cdv
+VCS_INFO_detect_cdv:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cdv:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cdv:9> VCS_INFO_check_com cdv
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cdv (/*)
+VCS_INFO_check_com:7> case cdv (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cdv:9> return 1
+vcs_info:68> vcs=cvs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:cvs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=cvs
+vcs_info:77> VCS_INFO_detect_cvs
+VCS_INFO_detect_cvs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_cvs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_cvs:9> VCS_INFO_check_com cvs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case cvs (/*)
+VCS_INFO_check_com:7> case cvs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_cvs:9> return 1
+vcs_info:68> vcs=darcs
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:darcs:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=darcs
+vcs_info:77> VCS_INFO_detect_darcs
+VCS_INFO_detect_darcs:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_darcs:7> [[ '' == --flavours ]]
+VCS_INFO_detect_darcs:9> VCS_INFO_check_com darcs
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case darcs (/*)
+VCS_INFO_check_com:7> case darcs (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_darcs:9> return 1
+vcs_info:68> vcs=fossil
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:fossil:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=fossil
+vcs_info:77> VCS_INFO_detect_fossil
+VCS_INFO_detect_fossil:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_fossil:7> [[ '' == --flavours ]]
+VCS_INFO_detect_fossil:9> VCS_INFO_check_com fossil
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case fossil (/*)
+VCS_INFO_check_com:7> case fossil (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_fossil:9> return 1
+vcs_info:68> vcs=git
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:git:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=git
+vcs_info:77> VCS_INFO_detect_git
+VCS_INFO_detect_git:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_git:7> [[ '' == --flavours ]]
+VCS_INFO_detect_git:9> VCS_INFO_check_com git
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case git (/*)
+VCS_INFO_check_com:7> case git (*)
+VCS_INFO_check_com:12> (( 1 ))
+VCS_INFO_check_com:12> return 0
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=+VCS_INFO_detect_git:9> git rev-parse --git-dir
+VCS_INFO_detect_git:9> vcs_comm[gitdir]=''
+VCS_INFO_detect_git:15> return 1
+vcs_info:68> vcs=hg
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:hg:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=hg
+vcs_info:77> VCS_INFO_detect_hg
+VCS_INFO_detect_hg:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_hg:7> [[ '' == --flavours ]]
+VCS_INFO_detect_hg:9> VCS_INFO_check_com hg
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case hg (/*)
+VCS_INFO_check_com:7> case hg (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_hg:9> return 1
+vcs_info:68> vcs=mtn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:mtn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=mtn
+vcs_info:77> VCS_INFO_detect_mtn
+VCS_INFO_detect_mtn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_mtn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_mtn:9> VCS_INFO_check_com mtn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case mtn (/*)
+VCS_INFO_check_com:7> case mtn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_mtn:9> return 1
+vcs_info:68> vcs=p4
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:p4:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=p4
+vcs_info:77> VCS_INFO_detect_p4
+VCS_INFO_detect_p4:1> local serverport p4where
+VCS_INFO_detect_p4:3> zstyle -t :vcs_info:p4:default:-all- use-server
+VCS_INFO_detect_p4:23> [[ -n '' ]]
+VCS_INFO_detect_p4:23> return 1
+vcs_info:68> vcs=svk
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svk:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svk
+vcs_info:77> VCS_INFO_detect_svk
+VCS_INFO_detect_svk:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svk:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svk:16> setopt localoptions noksharrays extendedglob
+VCS_INFO_detect_svk:17> local -i fhash
+VCS_INFO_detect_svk:18> fhash=0
+VCS_INFO_detect_svk:20> VCS_INFO_check_com svk
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svk (/*)
+VCS_INFO_check_com:7> case svk (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svk:20> return 1
+vcs_info:68> vcs=svn
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:svn:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=svn
+vcs_info:77> VCS_INFO_detect_svn
+VCS_INFO_detect_svn:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_svn:7> [[ '' == --flavours ]]
+VCS_INFO_detect_svn:9> VCS_INFO_check_com svn
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case svn (/*)
+VCS_INFO_check_com:7> case svn (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_svn:9> return 1
+vcs_info:68> vcs=tla
+vcs_info:69> [[ -n '' ]]
+vcs_info:70> (( 1 == 0 ))
+vcs_info:75> vcs_comm=( )
+vcs_info:76> VCS_INFO_get_cmd
+VCS_INFO_get_cmd:4> local cmd
+VCS_INFO_get_cmd:5> zstyle -s :vcs_info:tla:default:-all- command cmd
+VCS_INFO_get_cmd:6> vcs_comm[cmd]=tla
+vcs_info:77> VCS_INFO_detect_tla
+VCS_INFO_detect_tla:5> setopt localoptions NO_shwordsplit
+VCS_INFO_detect_tla:7> [[ '' == --flavours ]]
+VCS_INFO_detect_tla:9> VCS_INFO_check_com tla
+VCS_INFO_check_com:5> setopt localoptions NO_shwordsplit
+VCS_INFO_check_com:7> case tla (/*)
+VCS_INFO_check_com:7> case tla (*)
+VCS_INFO_check_com:12> (( 0 ))
+VCS_INFO_check_com:15> return 1
+VCS_INFO_detect_tla:9> return 1
+vcs_info:80> (( found == 0 ))
+vcs_info:81> vcs=-quilt-
+vcs_info:81> quiltmode=standalone
+vcs_info:82> VCS_INFO_quilt standalone
+VCS_INFO_quilt:3> (( 1 ))
+VCS_INFO_quilt:24> (( 1 ))
+VCS_INFO_quilt:63> (( 1 ))
+VCS_INFO_quilt:86> (( 1 ))
+VCS_INFO_quilt:92> emulate -L zsh
+VCS_INFO_quilt:93> setopt extendedglob
+VCS_INFO_quilt:94> local mode=standalone
+VCS_INFO_quilt:95> local patches pc qstring root
+VCS_INFO_quilt:96> local -i ret
+VCS_INFO_quilt:97> local context
+VCS_INFO_quilt:98> local -a applied unapplied applied_string unapplied_string quiltcommand quilt_env
+VCS_INFO_quilt:99> local -A hook_com
+VCS_INFO_quilt:101> context=:vcs_info:-quilt-.quilt-standalone:default:-all-
+VCS_INFO_quilt:102> zstyle -t :vcs_info:-quilt-.quilt-standalone:default:-all- use-quilt
+VCS_INFO_quilt:102> return 1
+vcs_info:82> VCS_INFO_set --nvcs
+VCS_INFO_set:7> setopt localoptions noksharrays NO_shwordsplit unset
+VCS_INFO_set:8> local -i i j
+VCS_INFO_set:10> [[ --nvcs == --nvcs ]]
+VCS_INFO_set:11> [[ '' == -preinit- ]]
+VCS_INFO_set:12> i=0
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_0_='
+VCS_INFO_set:12> i=1
+VCS_INFO_set:13> typeset -g 'vcs_info_msg_1_='
+VCS_INFO_set:15> VCS_INFO_nvcsformats
+VCS_INFO_nvcsformats:5> setopt localoptions noksharrays NO_shwordsplit
+VCS_INFO_nvcsformats:6> local c v rr
+VCS_INFO_nvcsformats:8> [[ '' == -preinit- ]]
+VCS_INFO_nvcsformats:13> zstyle -a :vcs_info:-quilt-:default:-all- nvcsformats msgs
+VCS_INFO_nvcsformats:14> (( 0 > maxexports ))
+VCS_INFO_nvcsformats:15> return 0
+VCS_INFO_set:16> [[ '' != -preinit- ]]
+VCS_INFO_set:16> VCS_INFO_hook no-vcs
+VCS_INFO_hook:5> local hook static func
+VCS_INFO_hook:6> local context hook_name
+VCS_INFO_hook:7> local -i ret
+VCS_INFO_hook:8> local -a hooks tmp
+VCS_INFO_hook:9> local -i debug
+VCS_INFO_hook:11> ret=0
+VCS_INFO_hook:12> hook_name=no-vcs
+VCS_INFO_hook:13> shift
+VCS_INFO_hook:14> context=:vcs_info:-quilt-+no-vcs:default:-all-
+VCS_INFO_hook:15> static=:vcs_info-static_hooks:no-vcs
+VCS_INFO_hook:17> zstyle -t :vcs_info:-quilt-+no-vcs:default:-all- debug
+VCS_INFO_hook:17> debug=0
+VCS_INFO_hook:18> (( debug ))
+VCS_INFO_hook:24> zstyle -a :vcs_info-static_hooks:no-vcs hooks hooks
+VCS_INFO_hook:25> (( debug ))
+VCS_INFO_hook:28> zstyle -a :vcs_info:-quilt-+no-vcs:default:-all- hooks tmp
+VCS_INFO_hook:29> (( debug ))
+VCS_INFO_hook:32> hooks+=( )
+VCS_INFO_hook:33> (( 0 == 0 ))
+VCS_INFO_hook:33> return 0
+VCS_INFO_set:19> (( 0 - 1 < 0 ))
+VCS_INFO_set:19> return 0
+vcs_info:83> return 0
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ set +x
+-zsh:67> set +x
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-11
[2026-05-13 11:40:52] ========================================
[2026-05-13 11:40:52] Screenpipe sync starting for: 2026-05-11
[2026-05-13 11:40:52] ========================================
[+00m00s] ▶ Preflight checks
Source DB: OK (4.5G)
NAS mount: OK /Volumes/screenpipe
[2026-05-13 11:40:53] Date 2026-05-11 already has 6857 frames in archive — skipping DB sync
Data dir: OK (283 files, 318M)
[+00m01s] ▶ Copying data folder for 2026-05-11
rsync 2026-05-11/ → NAS ✓ 0m01s (283 files, 318M)
[+00m02s] ▶ Copying audio files for 2026-05-11
rsync audio files → NAS skipped (no audio files for date)
[+00m02s] ▶ Copying screenpipe logs for 2026-05-11
rsync logs → NAS ✓ 1 file(s), 520K
[2026-05-13 11:40:54] Archive DB size: 1.3G
[2026-05-13 11:40:54] Total time: 0m2s
[2026-05-13 11:40:54] Sync complete for 2026-05-11
[2026-05-13 11:40:54] ========================================
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12
[2026-05-13 11:41:34] ========================================
UW PICO 5.09 New Buffer
[ Read 297 lines ]
^G Get Help ^O WriteOut ^R Read File ^Y Prev Pg ^K Cut Text ^C Cur Pos
^X Exit ^J Justify ^W Where is ^V Next Pg ^U UnCut Text ^T To Spell
UW PICO 5.09 File: screenpipe_sync_db.sh
Pico Help Text
Pico is designed to be a simple, easy-to-use text editor with a
layout very similar to the Alpine mailer. The status line at the
top of the display shows pico's version, the current file being
edited and whether or not there are outstanding modifications
that have not been saved. The third line from the bottom is used
to report informational messages and for additional command input.
The bottom two lines list the available editing commands.
Each character typed is automatically inserted into the buffer
at the current cursor position. Editing commands and cursor
movement (besides arrow keys) are given to pico by typing
special control-key sequences. A caret, '^', is used to denote
the control key, sometimes marked "CTRL", so the CTRL-q key
combination is written as ^Q.
The following functions are available in pico (where applicable,
corresponding function key commands are in parentheses).
^G (F1) Display this help text.
^F move Forward a character.
^B move Backward a character.
^P move to the Previous line.
^N move to the Next line.
^A move to the beginning of the current line.
^E move to the End of the current line.
^V (F8) move forward a page of text.
^Y (F7) move backward a page of text.
^W (F6) Search for (where is) text, neglecting case.
^L Refresh the display.
^D Delete the character at the cursor position.
^^ Mark cursor position as beginning of selected text.
Note: Setting mark when already set unselects text.
^K (F9) Cut selected text (displayed in inverse characters).
Note: The selected text's boundary on the cursor side
ends at the left edge of the cursor. So, with
^X Exit Help ^V Next Pg
[2026-05-13 11:41:34] Screenpipe sync starting for: 2026-05-12
[2026-05-13 11:41:34] ========================================
[+00m00s] ▶ Preflight checks
Source DB: OK (4.5G)
NAS mount: OK /Volumes/screenpipe
Archive DB: exists (1.3G)
Data dir: OK (220 files, 303M)
[+00m00s] ▶ Counting source rows for 2026-05-12
frames: 7274
elements: 853406
ui_events: 7044
ocr_text: 2306
meetings: 3
audio_chunks: 1113
audio_transcriptions: 319
[+00m01s] ▶ Initialising tables, indexes, FTS
creating tables ✓ 0m00s
creating indexes ✓ 0m00s
creating FTS tables ✓ 0m00s
[+00m01s] ▶ Syncing vision data for 2026-05-12
video_chunks ✓ 0m01s
frames (7274 rows) ✓ 1m05s
ocr_text (2306 rows) ✓ 0m49s
ui_events (7044 rows) ✓ 0m01s
elements (853406 rows) ✓ 0m46s
meetings (3 rows) ✓ 0m00s
[+02m43s] ▶ Syncing audio data for 2026-05-12
audio_chunks (1113 rows) ✓ 0m00s
audio_transcriptions (319 rows) ✓ 0m01s
[+02m44s] ▶ Updating FTS indexes
elements_fts ✓ 0m50s
frames_fts ✓ 1m13s
ui_events_fts ✓ 0m01s
audio_transcriptions_fts ✓ 0m01s
[+04m49s] ▶ Verifying DB
/Users/lukas/.screenpipe/scripts/lib/screenpipe_sync_db.sh: line 220: target_DATE: unbound variable
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ nano screenpipe_sync_db.sh
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $ ~/.screenpipe/scripts/screenpipe_sync.sh 2026-05-12
[2026-05-13 11:53:26] ========================================
[2026-05-13 11:53:26] Screenpipe sync starting for: 2026-05-12
[2026-05-13 11:53:26] ========================================
[+00m00s] ▶ Preflight checks
Source DB: OK (4.6G)
NAS mount: OK /Volumes/screenpipe
Archive DB: will be created
Data dir: OK (220 files, 303M)
[+00m00s] ▶ Counting source rows for 2026-05-12
frames: 7274
elements: 853406
ui_events: 7044
ocr_text: 2306
meetings: 3
audio_chunks: 1113
audio_transcriptions: 319
[+00m00s] ▶ Initialising tables, indexes, FTS
creating tables ✓ 0m00s
creating indexes ✓ 0m01s
creating FTS tables ✓ 0m00s
[+00m02s] ▶ Syncing vision data for 2026-05-12
video_chunks ✓ 0m00s
frames (7274 rows) ✓ 1m03s
ocr_text (2306 rows) ✓ 0m48s
ui_events (7044 rows) ✓ 0m01s
elements (853406 rows) ✓ 0m46s
meetings (3 rows) ✓ 0m00s
[+02m40s] ▶ Syncing audio data for 2026-05-12
audio_chunks (1113 rows) ✓ 0m00s
audio_transcriptions (319 rows) ✓ 0m00s
[+02m40s] ▶ Updating FTS indexes
elements_fts ✓ 0m32s
frames_fts ✓ 1m08s
ui_events_fts ✓ 0m00s
audio_transcriptions_fts ✓ 0m00s
[+04m20s] ▶ Verifying DB
frames: 7274 / 7274 ✓
elements: 853406 / 853406 ✓
ui_events: 7044 / 7044 ✓
ocr_text: 2306 / 2306 ✓
meetings: 3 / 3 ✓
audio_chunks: 1113 / 1113 ✓
audio_transcriptions: 319 / 319 ✓
[+04m29s] ▶ Copying data folder for 2026-05-12
rsync 2026-05-12/ → NAS ✓ 0m20s (220 files, 295M)
[+04m49s] ▶ Copying audio files for 2026-05-12
rsync audio files → NAS skipped (no audio files for date)
[+04m49s] ▶ Copying screenpipe logs for 2026-05-12
rsync logs → NAS ✓ 1 file(s), 304K
[2026-05-13 11:58:15] Archive DB size: 763M
[2026-05-13 11:58:15] Total time: 4m49s
[2026-05-13 11:58:15] Sync complete for 2026-05-12
[2026-05-13 11:58:15] ========================================
lukas@Lukas-Kovaliks-MacBook-Pro-Jiminny ~/.screenpipe/scripts/lib $
DOCKER
Close Tab
DEV (-zsh)
Close Tab
APP (-zsh)
Close Tab
-zsh
Close Tab
screenpipe"
Close Tab
⌥⌘1
-zsh...
|
NULL
|
NULL
|
NULL
|
NULL
|