Skip to content

Transcribe recordings

Transcribe writes down what’s said in a recording.

  1. Install a transcription model from Get models → Transcribe. Whisper models are a good choice: the small ones are fast, the large ones are more accurate.
  2. Open Transcribe and either upload an audio file or record from your microphone.
  3. Press Write it down.
Option What it does
Language Detect automatically, or pick the spoken language for better accuracy.
Translate to English Writes the transcript in English, whatever language is spoken. Supported by Whisper models.
  • Copy puts the text on your clipboard.
  • Save as text downloads transcript.txt.
  • Save as subtitles downloads subtitles.srt, ready for video players and editors. This appears when the model provides timestamps, which Whisper models do.

The same transcription model powers the microphone button in Chat.