Runs in the browser · Text to speech
Run Supertonic TTS 2 in your browser
Supertonic TTS 2 downloads 250.7 MB over WebGPU (fp32), or 250.7 MB over WebAssembly (fp32), to turn text into a spoken voice entirely in the tab with Transformers.js. No install, no server.
Text to speech · onnx-community/Supertonic-TTS-2-ONNX
- Parameters
- -
- Pipeline task
- custom
Reading
Ships as three separate ONNX parts (text encoder, latent denoiser, voice decoder) in fp32 only, no quantized build published yet. There is no pipeline() task that wires a 3-stage TTS graph together; expect custom ONNX Runtime code.
Will it run in your browser?
Checking for WebGPU support…
All measured variants
| Variant | Size |
|---|---|
| fp32 WebGPU pick WASM pick | 250.7 MB |
Sizes measured from the HuggingFace API file tree for onnx-community/Supertonic-TTS-2-ONNX, not estimated.
Use it with Transformers.js
There is no stable transformers.js pipeline() call for this model yet, and no verified low-level snippet either (see the Reading panel above). Check the model card on HuggingFace for current usage examples.
Sources
Same job, different size
FAQ
WebGPU or WASM for Supertonic TTS 2, in practice?
Chrome, Edge, and Safari 26+ run the WebGPU build (fp32) on the GPU. Firefox has WebGPU on Windows and Apple Silicon Macs but not everywhere yet. Any browser without a WebGPU adapter falls back automatically to the WebAssembly build (fp32) on CPU: same model, slower to load, slower to run.
What does fp32 mean for Supertonic TTS 2?
The WebGPU build here uses fp32: full 32-bit floating point, the uncompressed original weights. The WASM fallback uses fp32: full 32-bit floating point, the uncompressed original weights.
Sizes measured 2026-08-01 from the HuggingFace API. Last validated 2026-08-03. See all browser models or the methodology.