Choose a transcription model
Outcome
Section titled “Outcome”New imports, recordings, and voice-input workflows have a selected local Whisper model.
How to choose
Section titled “How to choose”- Open Models from the sidebar or press Cmd+M.
- In Transcription, select an installed model to make it the default.
- Select an uninstalled catalog model to download it. A missing-model dialog can also download the model required by the workflow you started.
- Test a representative non-sensitive recording before changing a team workflow around that model.
Current catalog
Section titled “Current catalog”| Capability | Status | Current behavior |
|---|---|---|
| Whisper Tiny | Available | Smallest Whisper model exposed by the in-app transcription catalog. |
| Whisper Base | Available | Lightweight Whisper model exposed by the in-app transcription catalog. |
| Whisper Small | Available | Whisper model recommended by the current heuristic for many lower-memory machines. |
| Whisper Medium | Available | Whisper model used by the current recommendation heuristic for Intel machines with sufficient memory. |
| Whisper Large v3 Turbo | Available | Default transcription model and the current Apple Silicon recommendation for many machines with at least 16 GB of memory. |
| Whisper Large v3 | Available | Largest Whisper model exposed by the in-app transcription catalog. |
The table’s sizes are catalog estimates. They do not establish a disk-space requirement, download guarantee, processing speed, or transcription-accuracy guarantee.
What happens when a model is missing
Section titled “What happens when a model is missing”Synthezia stops the action at a model-requirement dialog rather than attempting to transcribe without a model. Select Download and continue or cancel the workflow; the dialog also displays the current estimated download size.