audio.cpp again. Hopefully you are not sick of it yet :) Release 0.3 adds five new models: Supertonic 3, MOSS-TTS-Local, MOSS-TTS-Nano, IndexTTS2, and Irodori-TTS. The highlight is Supertonic 3. It can hit 200×+ real time on CUDA (RTX 5090), 6×+ on CPU, and around 47 ms TTFT in CUDA streaming mode.