ChatTTS

Turning text into natural-sounding spoken dialogue in Chinese and English is the job ChatTTS does. ChatTTS is a text-to-speech model built for conversational scenarios, so it suits the voice of a large language model assistant better than it suits long-form narration.

You can also use it for conversational audio and video introductions.

The model was trained on roughly 100,000 hours of Chinese and English data, which is where its speech quality and naturalness come from.

The project team has said it plans to open-source a base model trained on 40,000 hours of data, geared toward academic researchers and developers who want to build on it.

That makes ChatTTS most relevant if you work with open source AI models and want a voice layer for a chatbot or an API-driven app.

Language support is limited to the two languages above.