Voice licences

We only offer voices whose licence allows commercial use. Some licences require crediting the dataset authors; those credits are listed below. The voice models run in your browser; we do not claim ownership of them.

Attribution

Loading…

Open-source components

• Piper voices: Piper (MIT) by Rhasspy / Open Home Foundation.
• Kokoro voices: Kokoro-82M (Apache-2.0) by hexgrad; includes audio under CC BY (SIWIS, Koniwa).
• Speech-to-phoneme conversion uses espeak-ng (GPL-3.0), compiled to WebAssembly; its source is available at that link.
• Voice cloning: Kyutai Pocket TTS (CC BY 4.0), ONNX export by KevinAHM (CC BY 4.0).
• Voice cloning: Pocket TTS Hindi by Saryps Labs (CC BY 4.0), based on Kyutai Pocket TTS and trained on AI4Bharat IndicVoices and Kathbath (CC BY 4.0).
• onnxruntime-web (MIT) and transformers.js (Apache-2.0) run the models.