AI · GitHub repo
Supertonic
Fast, lightweight on-device text-to-speech running locally via ONNX Runtime.
Supertonic is an open-source, on-device text-to-speech model built to execute fast local inference without relying on cloud APIs or dedicated GPUs. Operating via ONNX Runtime with a compact 99M-parameter footprint, it produces 44.1kHz audio across 31 languages on hardware ranging from desktop CPUs to Raspberry Pis. Note that the original repository has been archived, though code and weights remain usable under the MIT license.
What it's for
- Run offline, low-latency text-to-speech synthesis directly on user devices without API costs.
- Convert long web articles into spoken audio inside browser extensions in under a second.
- Add speech playback to resource-constrained devices like Raspberry Pi boards or e-readers.
- Integrate multilingual speech synthesis across mobile apps using iOS, Flutter, or native SDKs.
Features
- Lightweight ONNX architecture requiring no GPU and running entirely offline.
- Multilingual support covering 31 languages, including a language-agnostic auto-detection mode.
- High-quality audio generation directly outputting 44.1kHz 16-bit WAV files.
- Native handling of complex numbers, currency, dates, and technical abbreviations.
- Inline expression tags like laughter, breaths, and sighs to adjust vocal nuance.
- Broad language bindings including Python, Node.js, WebGPU, Rust, Go, C++, C#, Java, Swift, and Flutter.
Pricing
Open source. Free and open-source under the MIT license.
Details
- Type
- GitHub repo
- Category
- AI
- Topic
- Text to Speech
- Open source
- Yes
- Platforms
- macOS, Windows, Linux, iOS, Android, Web, Browser extension, Code library, Command line
- For
- Developers building offline, edge-native, or privacy-focused audio applications.
- Alternative to
- ElevenLabs, OpenAI TTS
- Tags
- tts, speech synthesis, onnx, on device, voice, audio
Why it was saved
- @marc.kaz: Eliminates cloud TTS API costs and latency, running 167x faster than real-time on Apple Silicon and operating efficiently even on low-power devices like Raspberry Pi and e-readers.
Compare
Related
AI voice generator, voice cloning, and conversational agent platform
Ultra-low-latency streaming text-to-speech and instant voice cloning model.
Open-source 1.6B text-to-speech model generating multi-speaker dialogue in one pass.
Fast, low-cost text-to-speech API and studio for creators and developers
Converts spoken thoughts into structured, polished writing in customizable styles.
Official Python library for Cartesia's real-time text-to-speech API
Updated