XTTS v2
A local text-to-speech model that clones voices from short recordings and generates multilingual speech through Coqui TTS.
Alternative to:
XTTS v2 generates speech from text on your own hardware, using a voice recording as a reference. It's for developers adding speech to applications and researchers adapting a speech model to their own data.
- Voice cloning: Generate speech from a six-second reference clip, carrying over the speaker's emotion and speaking style.
- Cross-language speech: Use a reference voice to speak another supported language, including English, Spanish, Arabic, Chinese, Japanese and Hindi.
- Reusable voices: Combine multiple reference recordings, blend speakers, and save cloned voices in Coqui TTS for reuse without supplying the audio each time.
- Fine-tuning: Adapt the model to your own training data with Coqui TTS's fine-tuning code.
Coqui TTS provides Python and command-line access, plus a self-hosted speech server and Docker images. You can save generated speech as WAV audio. The toolkit uses PyTorch and supports NVIDIA CUDA execution. It runs on Ubuntu and is expected to work on macOS and Windows.
The idiap/coqui-ai-TTS toolkit is a fork of coqui-ai/TTS. Its open-source library code uses Mozilla Public License 2.0; the XTTS v2 model weights use the separate Coqui Public Model License.
