MegaTTS3 is a PyTorch implementation for text-to-speech (TTS) using a lightweight and efficient diffusion transformer model. It offers high-quality voice cloning, bilingual support, and controllable accent and pronunciation features.
