Supertonic

View Poster Visit Website Search Youtube Videos

Supertonic Overview

Supertonic is a lightning-fast, on-device multilingual text-to-speech (TTS) system from Supertone Inc., designed for local inference with minimal overhead. Powered by ONNX Runtime, it runs entirely on your device — no cloud, no API calls, no privacy concerns. Supertonic 3 is the latest model generation, expanding to 31 languages with improved reading accuracy and fewer repeat/skip failures.

Service Notice: As of July 23, 2026 the repository is archived — there will be no further development or official support for the open-source models, and Voice Builder will no longer be accessible after August 31, 2026.
Supertonic 3: The current generation adds 31-language support, improved reading accuracy, fewer repeat/skip failures, and a v2-compatible public ONNX interface in a ~99M-parameter open-weight model.
Blazingly Fast: Low-latency, real-time synthesis across desktop, browser, mobile, and edge — fast enough to turn an entire webpage into audio in under a second.
31-Language Multilingual: Synthesize speech directly from text across 31 languages, or pass lang="na" for language-agnostic input when you don't know the source language.
Local HTTP Server: The Python SDK (pip install supertonic) can run supertonic serve — a local HTTP server with native /v1/tts and OpenAI-compatible /v1/audio/speech endpoints for local agents, browser extensions, and automation tools.
Edge-Device Ready: Runs locally on desktop, mobile, browsers, and resource-constrained hardware like Raspberry Pi or e-readers (0.3× real-time), with zero network dependency and no GPU required.
44.1kHz High-Quality Audio: Outputs studio-grade 44.1kHz 16-bit WAV directly, ready for production playback without any external upsampler.
Expression Tags: 10 inline tags (e.g. <laugh>, <breath>, <sigh>) bring natural human nuance into generated speech without prompt engineering or reference audio.
Multi-Runtime SDKs: Ready-to-use ONNX Runtime examples across Python, Node.js, Browser (WebGPU), Java, C++, C#, Go, Swift, iOS, Rust, and Flutter.
Voice Profiles & Cloning: The open-weight repo focuses on fixed-voice TTS; Voice Builder turns a short reference recording into version-specific JSON voice profiles for Supertonic 2 and 3, while Supertone Play and the Supertone API offer hosted workflows with zero-shot voice cloning.
Open Weight & License: A ~99M-parameter fully open-weight checkpoint — a fraction of the size of 0.7B to 2B class open TTS systems — released under MIT (sample code) and OpenRAIL-M (model).