Kitten TTS
lastmod 2025-08-26created 2025-08-072 edits
A 25MB open-source TTS model that runs on CPU alone
About
- Kitten TTS is an open-source TTS (text-to-speech) model that aims for both lightweight size and high audio quality
- Uses only 15 million parameters, with a model size under 25MB
- Its key feature is that, unlike other large TTS models, it can run in any environment, including mobile and embedded
Capable of high-quality speech synthesis on any device without a GPU Offers a variety of premium voice options, supporting high-quality speech synthesis close to real voices Fast speech inference, optimized for real-time synthesis A developer preview model has been released; the fully trained model weights, a mobile SDK, a web version, and more are planned for sequential release See also
Favorite site