Chatterbox
SoTA open-source TTS under the MIT license, with voice cloning support
A SoTA open-source TTS under the MIT license, supporting voice cloning.
Text-to-speech, speech-to-text, and voice cloning, open source.
Key features
- Family of open-source TTS models by Resemble AI
- Multilingual V3: 23+ languages with better speaker similarity
- Turbo: 350M low-latency model for voice agents
- Nano: 110M model for on-device CPU inference, 3x realtime
- Native paralinguistic tags for laughs, coughs, and realism
Good for
- Low-latency voice agents and narration
- Cross-language voice cloning and localization
Stack & tags
Pythonttstext-to-speechvoice-cloningmultilingualvoice-agentson-deviceopen-weights
Quick start
pip install chatterbox-ttsFrom the project README on GitHub.
Similar open-source projects
Whisper
★ 109,795OpenAI's open-source speech recognition model — the benchmark for multilingual STT
GPT-SoVITS
★ 62,271TTS/cloning trainable on 1 minute of audio — a legendary Chinese-community project