Next-generation TTS model using flow-matching and DiT, inspired by Stable Diffusion 3 - View it on GitHub