Skip to content

Transcripts

Audio is segmented and run through FluidAudio’s neural ASR model on Apple silicon. Speakers are diarized per-channel: system-audio speakers and microphone speakers stay distinct.

A typical 1-hour recording transcribes in under 10 minutes on an M2.

Click any transcript line to edit it. Edits are saved locally and never overwrite the source audio.

Long-press a speaker label to see candidate matches from your voice profile directory. Click “Apply to all turns” to relabel every line attributed to that speaker.