YourTTS

YourTTS is a multilingual multi-speaker text-to-speech model developed by Edresson Casanova and collaborators at the University of São Paulo with Coqui, published on arXiv in December 2021. It builds on the VITS end-to-end architecture and adds a speaker encoder, enabling zero-shot multi-speaker synthesis and zero-shot voice conversion.

The model achieved state-of-the-art results in zero-shot multi-speaker TTS on VCTK and can be fine-tuned on less than one minute of speech to clone a new voice with high similarity. Because it can adapt to a new language from a single-speaker dataset, it opened practical TTS for low-resource languages. The release date below is era-level (2022), when the model shipped in the Coqui toolkit.

YourTTS shipped in the Coqui TTS toolkit as tts_models/multilingual/multi-dataset/your_tts and remains a widely used open baseline for voice cloning research.