Supertonic (Experimental)
Bundledby TypeWhisper
About
Supertonic is TypeWhisper’s experimental local text-to-speech provider for Supertonic 3 ONNX models. It downloads the model assets into TypeWhisper’s plugin data directory, then generates speech locally for spoken feedback and readback.
The model assets are licensed separately from TypeWhisper under OpenRAIL-M. TypeWhisper requires you to review and accept the Supertonic 3 model license before it downloads the model files.
The settings shown above document the current macOS plugin surface. The catalog entry also lists Windows so the add-on appears in Windows platform filtering once the matching Windows bundle is available.
Requirements & Privacy
| Requirement | Details |
|---|---|
| TypeWhisper | Requires TypeWhisper 1.4.0 or newer. |
| macOS | Requires macOS 14.0 or newer on Apple Silicon (arm64). |
| Model download | Downloads Supertonic 3 ONNX files and voice style JSON files from Hugging Face. |
| License gate | Model download is disabled until the OpenRAIL-M terms have been accepted in the settings view. |
| Network use | Network access is used for model downloads and optional Hugging Face token validation. Speech generation runs locally after the model is cached. |
| Credentials | No API key is required for synthesis. A Hugging Face token is optional and only helps with download rate limits. |
Model License & Download
| Field | What it controls | Notes |
|---|---|---|
| Open Supertonic 3 OpenRAIL-M license | Opens the model license before download. | Review the restrictions before accepting. |
| I have read and accept… | Stores acceptance for the current Supertonic 3 license id and revision. | Download remains disabled until this is checked. |
| Download & Load | Downloads required ONNX model files and voice style JSON files. | Progress is shown while the files are fetched. |
| Ready | Confirms the required model assets are present locally. | TypeWhisper can synthesize speech once this state appears. |
| Delete cached model | Removes the cached model directory. | Use it to free disk space or force a fresh download. |
| Hugging Face Token | Optional token used for Hugging Face requests. | Useful when unauthenticated downloads hit rate limits. |
Voice & Quality
| Setting | Values | Notes |
|---|---|---|
| Voice | Defaults to M1; bundled style names include M1 and F1, with additional discovered voice style files listed when present. | The plugin exposes available voice style JSON files as TypeWhisper voices. |
| Speed | 0.70 to 2.00, step 0.05; default 1.05. | Lower values slow the generated speech, higher values speed it up. |
| Quality | Fast, Balanced, High; default Balanced. | Higher quality uses more synthesis steps. |
Supported language hints include en, de, fr, es, it, ja, ko, pt, nl, pl, sv, tr, uk, vi, and more. Unsupported or empty language hints fall back to English.
Setup
- Open TypeWhisper Settings > Plugins.
- Configure Supertonic (Experimental).
- Review the Supertonic 3 OpenRAIL-M license.
- Accept the model license terms.
- Add a Hugging Face token only if downloads are rate-limited.
- Click Download & Load and wait for Ready.
- Choose a voice, speed, and quality mode.
- Select Supertonic as the text-to-speech provider for spoken feedback or readback.
Notes
- Supertonic is experimental and currently documented from the macOS settings surface.
- Speech generation is local after the model download completes.
- Model assets are downloaded separately because they are governed by the Supertonic 3 OpenRAIL-M license.
- Removing the cached model does not remove your license acceptance or optional Hugging Face token.