Coqui TTS (idiap fork)

Coqui TTS is a text-to-speech toolkit for developers and speech researchers who want to generate voices locally or train their own models. The idiap fork builds on coqui-ai/TTS.

• Speech models: Generate speech with pretrained or custom models, including XTTS, VITS, YourTTS, Tortoise and Bark, with multilingual and multiple-speaker support depending on the model.
• Voice cloning: Generate speech from a reference voice and save cloned voices for reuse; the toolkit can also combine a single-speaker model with FreeVC for voice cloning.
• Voice conversion: Change the voice in an existing recording to match a target recording using FreeVC, kNN-VC or OpenVoice.
• Training and fine-tuning: Train new speech models or adapt existing ones, with dataset analysis and curation tools and training logs available in TensorBoard.
• Application integration: Use Python or command-line interfaces for speech generation, with a lower-level synthesis API for additional outputs such as timestamps.

The open-source library uses Mozilla Public License 2.0. It runs locally or on a self-hosted server, with Docker images available. Linux is its tested platform; macOS and Windows are expected to work too. Its PyTorch backend supports CPU and NVIDIA CUDA execution.