Local AI Voice?
Fully Offline. Free.

Want natural AI voice generation running fully offline with zero API costs? You can do it right on your computer.

Typically, for high-quality text-to-speech, you call cloud APIs. This costs money per character, adds latency, and risks data leakage.

This is Supertonic by Supertone on GitHub. A multilingual text-to-speech system running directly on-device via ONNX Runtime.

Compared to OmniVoice Studio, which does heavy GPU voice cloning, Supertonic is a lightweight 99M parameter TTS engine running smoothly on a standard CPU.

It supports thirty-one languages, including English and Vietnamese. Simply pass lang='na', and the model auto-detects the language.

Thanks to its ultra-lightweight size, Supertonic runs well on Raspberry Pi, Onyx Boox e-readers, or as a Chrome Extension.

The latest update integrates a Local HTTP Server, creating an endpoint fully compatible with OpenAI's audio/speech API.

Supertonic is a secure, private, and free local TTS toolkit worth trying. Follow AI News twenty-four seven for more!