silero-models
snakers4/silero-models
Silero Models: pre-trained text-to-speech models made embarrassingly simple
text-to-speechspeech-synthesispretrained-modelspytorchtorch-hubrussianmultilingualspeechmachine-learningdeep-learning
Summary
Silero Models is a popular open-source repository providing pre-trained text-to-speech models for multiple languages, with a strong focus on Cyrillic languages (Russian, Ukrainian, Belarusian, etc.) and other regional languages like Armenian, Georgian, and Indic languages. The project emphasizes simplicity with one-line inference, fast CPU/GPU performance, and features like automated stress marking for Russian. It is distributed via PyTorch Hub and a pip package, and includes extensive documentation and Colab examples.