Privacy
Sonari's whole design is that your voice never leaves your Mac. This isn't a setting you switch on. There's no cloud to opt out of.
- Recording happens on your Mac.
- Transcription happens on your Mac, using a model you downloaded.
- Cleanup and AI transforms happen on your Mac, through Apple's on-device Foundation Models. If they're unavailable, your transcript is inserted unchanged and is never sent to a server.
- App matching (picking a transform based on where you're dictating) reads only which app is in front. Nothing else. That check stays on your Mac, needs no permission, and the app name is never sent anywhere.
There is no account to create, no API key to paste, and no cloud fallback.
When Sonari uses the network
Only to download a model you chose. Once a model is cached, dictation works fully offline: on a plane, off Wi-Fi, anywhere.
Where your data lives
Everything is stored under your own Library folder:
| What | Where |
|---|---|
| Recordings + transcript index | ~/Library/Application Support/Sonari/ |
| WhisperKit models | ~/Library/Application Support/Sonari/WhisperKit/ |
| Parakeet models | ~/Library/Application Support/FluidAudio/Models/ |
You can delete any recording and its audio from the History window whenever you like.