When you stop a session, Confidant immediately starts transcribing. Here's what to expect.
Stopping
Click Stop in the session panel. The audio file is closed and saved to disk, and the panel transitions to the transcription phase.
There's no "discard" option. If you stopped a recording you didn't want to keep, you can delete the session from the patient detail page after it finishes transcribing. To have audio cleared automatically going forward, set an audio retention policy in Settings → Privacy & Security → Privacy & Storage — the choices are Keep forever, Delete after transcribed (audio cleared as soon as transcription completes), and Delete after note is finalized (audio cleared when you finalize the note).
Transcription
The status indicator turns amber and reads Transcribing…. The transcription engine processes the audio in chunks, and you'll see a progress indicator like "12 / 47" — that's the number of audio chunks processed out of total chunks expected.
How long it takes depends on session length and your hardware:
Session length | M2/M3/M4 Mac | Recent Intel Mac / PC |
30 min | 5–8 min | 12–18 min |
50 min | 8–14 min | 20–30 min |
90 min | 15–25 min | 35–55 min |
You can keep using the rest of Confidant while transcription runs in the background. The session panel can be closed; you'll get a chime when it's done.
After transcription
The status indicator turns sage green and reads Transcribed. The session panel now shows the full transcript text. You can:
Read the transcript inline
Click Generate notes to have the AI write a draft SOAP or DAP note
Open the patient detail page to see the session in the patient's session list
The transcript is also saved to disk (encrypted, like everything else) and stays available regardless of whether you generate notes from it.
What if transcription fails or stalls?
Most transcription issues are CPU/RAM contention with other apps. Try:
Closing other audio/video apps (Zoom, OBS, video editors)
Closing Chrome tabs (especially tabs running video)
Quitting any large language model apps you have open separately
If it still won't progress, see Troubleshooting. Your audio is safe on disk regardless — even if transcription fails, you can re-trigger it.