Re-transcribe a meeting

To get a better transcript of a past call, open the meeting and click Re-transcribe. Fetch listens to the saved audio again, usually with a more careful model, and replaces the transcript. To see which model does best on your meetings first, click Compare models.

When should I re-transcribe?

• Names or jargon came out wrong.
• The meeting was in another language and the model didn’t catch it.
• The after-call clean-up was turned off, or the computer was too busy to finish it.
• You downloaded a newer or more accurate model and want older calls to benefit.

How do I re-transcribe?

1. Open the meeting.
2. Click Re-transcribe at the top.
3. Choose the model to use. Models you haven’t downloaded yet can be fetched from Settings > Transcription.
4. Wait while Fetch works. Long calls take a while on slower computers, so this is a good moment for a walk. Dogs recommend it.

When it’s done, the new transcript replaces the old one. Click Enhance notes to refresh the AI notes from it.

How does Compare models work?

Compare models runs two speech models on the same audio and shows their transcripts side by side, so you can judge which handles your voices, accents and vocabulary best. It’s the honest way to choose, because public test scores don’t know your team. Then set your favorite in Settings > Transcription (see Choose a speech model).

Do I need the audio?

Yes. Re-transcribe and Compare both listen to the recording. If you’ve set Fetch to delete audio (in Settings > Recording), older calls can’t be re-transcribed. See Where are my recordings saved? for the audio settings.

Does re-transcribing send my audio anywhere?

No, unless you pick the online Deepgram option. Every other model runs on your own computer.