A speech model turns your audio into words on your Mac. Install at least one before dictating. Writing cleanup uses a separate, optional model.
Models in this release
| Model | Live captions | Approximate download |
|---|---|---|
| Best Quality (default) | Yes, when enabled | 550 MB |
| Balanced | No; text after stopping | 1.7 GB |
| Fast | No; text after stopping | 550 MB |
All three are offered for English dictation in TessFlo. A model vendor’s wider language list does not describe TessFlo’s supported languages.
Allow extra space for installation and preparation. Download sizes are estimates, not final storage sizes. Model names do not guarantee speed or accuracy—try them with your own voice.
Download or switch
- Open Settings → Models. Expand the additional choices if needed.
- Select a model, choose Download, and review its notice.
- Stay connected while it downloads. Keep TessFlo open for Checking model… after the transfer reaches 100%.
- To switch to an installed model, use its selection action. You do not need to delete your working model first.
Downloaded means the files are installed. Preparing means TessFlo is getting the model ready to dictate. First use, or returning after a long idle period, may take longer. Follow the displayed Start or retry guidance.
If another model is downloading or being checked, TessFlo may offer an installed model so you can keep dictating. Changing models does not remove Stop or Cancel from an active recording.
A speech-model retry starts a fresh download. Light Cleanup handles interrupted downloads separately and may resume differently.
Enable captions
- Select Best Quality and finish its installation.
- Enable Settings → Recording → Show recording overlay.
- Enable Live captions and start a test recording.
Words appear in the overlay as you speak. They are a preview, not live insertion into another app. If captions cannot run, TessFlo can still capture audio and provide text after you stop. Balanced and Fast work without captions.