A Non-Autoregressive Transformer based Text-to-Speech, supporting a family of SOTA transformers with supervised and unsupervised duration modelings. This project grows with the research community, aiming to achieve the ultimate TTS
text-to-speech deep-learning unsupervised pytorch tts speech-synthesis transformer supervised multi-speaker sota comprehensive single-speaker neural-tts non-autoregressive fastspeech fastspeech2 hifi-gan non-ar mel-gan ultimate-tts
-
Updated
Sep 24, 2022 - Python