ChatTTS
StaleDescription
A generative speech model for daily dialogue, providing AI agents with natural and fluent voice synthesis with fine-grained prosody control.
Key Features
- Conversational TTS — Optimized for dialogue scenarios like LLM assistants with natural expressive synthesis
- Fine-grained prosody control — Predict and control laughter, pauses, interjections and subtle prosodic features
- Multi-speaker support — Sample random speaker timbre, supports multi-character dialogue scenarios
- Streaming audio generation — Real-time streaming voice output support
- Chinese and English support — Trained on 100K+ hours of Chinese and English audio data
- WebUI and CLI — Both web interface and command line usage modes available
Use Cases
💡 Add natural voice replies to AI conversation assistants
💡 Generate multi-character voice content for podcasts or audiobooks
💡 Integrate TTS into voice messaging applications
💡 Provide text-to-speech support for accessibility applications
💡 Build interactive voice IVR systems
Strengths & Limitations
✅ Strengths
- • High community interest (39.8k stars)
- • Established track record (2 years in production)
- • Responsive to issues, low backlog
⚠️ Limitations
- • Restrictive license (AGPL-3.0)
Categories
Quick Start
```bash
pip install ChatTTS
```
```python
import ChatTTS
import torch
import torchaudio
chat = ChatTTS.Chat()
chat.load(compile=False)
texts = ["Hello, I am ChatTTS.", "The weather is nice today."]
wavs = chat.infer(texts)
for i, wav in enumerate(wavs):
torchaudio.save(f"output{i}.wav", torch.from_numpy(wav), 24000)
```