Configuration Reference
All settings live under Settings, the gear icon at the bottom of the autorec window’s sidebar. Changes apply once you press Save Settings. Configuration is stored in ~/.config/autorec/config.json on Linux and macOS, and %APPDATA%\autorec\config.json on Windows.
General
| Setting | Default | Description |
|---|---|---|
| Launch Autorec at login | On (enabled on first run) | Starts autorec hidden in the system tray when you sign in |
Output
| Setting | Default | Description |
|---|---|---|
| Output Directory | ~/Videos/autorec (your Videos folder on Windows) | Where recordings are saved |
| Filename Template | {datetime}_{app} | Pattern for recording filenames. Tokens: {datetime} (YYYY-MM-DD_HH-MM) and {app} |
Recording
| Setting | Default | Description |
|---|---|---|
| Quality | High (1080p) | High (1080p), Medium (720p), or Low (480p) |
| Framerate | 30 | Frames per second: 15, 30, or 60 |
There is no encoder setting. Autorec tries the hardware H.264 encoders your machine offers (NVENC, Quick Sync, VideoToolbox on macOS, Media Foundation on Windows, VAAPI) and falls back to software OpenH264 when none of them opens. Hardware encoding uses less CPU.
Detection
| Setting | Default | Description |
|---|---|---|
| Zoom | Enabled | Detect and record Zoom meetings |
| Microsoft Teams | Enabled | Detect and record Teams meetings, in the desktop app and in a browser |
| Google Meet | Enabled | Detect and record Google Meet meetings |
| Proton Meet | Enabled | Detect and record Proton Meet calls, in a browser and in the desktop app |
| Jitsi Meet | Enabled | Detect and record calls on meet.jit.si |
| Signal | Enabled | Record Signal calls (starts when Signal opens the microphone) |
| Element | Enabled | Record Element calls (starts when Element opens the microphone) |
| Telegram | Enabled | Record Telegram calls (starts when Telegram opens the microphone) |
Autorec checks for meetings every 5 seconds. Each entry under Settings → Watchers can be enabled, edited or removed, and Restore Defaults brings deleted ones back. A disabled watcher detects and records nothing.
Adding your own app
Any other call app works too: Discord, Webex, a self-hosted Jitsi. Open Settings → Watchers → Add an app, pick the app from the list of running programs, and choose when to record:
- The window title looks like a meeting suits apps whose title changes during a call, like Zoom or Jitsi. Autorec suggests a pattern from the current title.
- The app is using the microphone suits apps whose title never changes, like Telegram or Signal. Recording starts when the app opens the microphone and stops shortly after it lets go.
- The app is running at all records whenever the app is open. Fine for an app you only launch for calls, noisy for one that stays open all day.
If the app isn’t open right now, use Browse for the app file instead. The watcher shows a “not running now” badge until autorec sees the app for the first time.
Transcription
| Setting | Default | Description |
|---|---|---|
| Auto-transcribe recordings | Off | Transcribe each recording when it finishes |
| Model | base | Whisper model size: tiny, base, small, medium, large |
| Generate SRT subtitles | On | Write an .srt file next to the .txt transcript |
| Keep intermediate audio (WAV) | Off | Keep the extracted audio file after transcription |
| Force CPU | Off | Skip the GPU and transcribe on the CPU |
Models are not downloaded automatically. Download one in Settings → Transcription before you turn on auto-transcribe; without the model file, auto-transcribe does nothing. Larger models produce more accurate transcriptions but take longer to process. See Transcription & AI for details on model sizes.
AI Summary
| Setting | Default | Description |
|---|---|---|
| Provider | — | A preset that fills in the base URL and suggests models, or Custom |
| API Base URL | — | URL of any OpenAI-compatible API |
| API Key | — | Your API key for the AI service |
| Model | gpt-4o-mini | Which model to use |
| Summary prompt | Built in | The system prompt sent to the model. Edit it to ask for decisions, action items or whatever you need |
| Max summary length (tokens) | 1000 | 300 is a short paragraph, 1000 is roughly a page |
| Auto-summarize after transcription | Off | Generate a title and summary from each new transcript |
AI summaries require an API key from any OpenAI-compatible service (OpenAI, OpenRouter, etc.). Autorec sends only the transcript text — no audio or video data leaves your machine.