Transcribe
What it does
Turns audio and video into text. Cloud mode uses the speech-to-text provider you configured. Local mode uses whisper.cpp on this machine.
Setup
Bring your own STT key (or install whisper.cpp for on-device). Cadence does not bill transcription in 0.1. Large files are split around the 25MB boundary. Video has audio extracted automatically.
Permissions
- A configured STT key, or a local whisper.cpp install.
- File access to the media.
Common prompts
| You say | What happens |
|---|---|
| “Transcribe this meeting recording” | Audio to text |
| “Transcribe this video locally” | On-device whisper.cpp |
| “Transcribe this podcast and summarize it” | Transcript plus a summary |
Tips
- Local mode keeps audio on disk. Cloud mode is faster if you have a key.
- Turn a transcript into a long draft with Document.