ChatTTS

ChatTTS is a voice generation model purpose-built for conversational scenarios, delivering some of the most natural and high-quality speech synthesis available for Chinese and English. If you've tried other text-to-speech tools and been left with robotic, unnatural output that breaks the immersion for your users — ChatTTS is a step up. It's ideal for dialogue-driven applications like large language model assistant interfaces, as well as conversational audio and video content. The model's exceptional quality comes from training on approximately 100,000 hours of Chinese and English speech data. The project team also plans to open-source a foundational model trained on 40,000 hours of data, offering the academic and developer community a powerful resource for further research and development.