Prepare transcription
- Open Meetings or Record a Meeting from the available Tools or chat actions.
- Check the transcription engine in Settings. Download and prepare a local Whisper model if you use Whisper.
- Grant microphone access for recording and speech-recognition permission when the selected engine requests it.
- Enable speaker separation if needed and prepare any speaker model Noema requests before recording.
Record or import
Record a conversation or import an existing audio recording. Noema transcribes successive audio chunks and places their timestamps on the meeting timeline.
Review the transcript alongside playback. When speaker separation is enabled, label the speakers and correct uncertain attribution before relying on who said what.
Bring the transcript into chat
Use the meeting’s chat action to bring its transcript into a conversation. Ask for a summary, decisions, action items, or answers tied to what was said.
For local transcription, use Whisper or require on-device Apple Speech recognition. A chat using a remote model can send the transcript to that provider. Choose a local model when the subsequent analysis should remain local.