⚡ Non-autoregressive TTS using Conditional Flow Matching - 5-20x faster than AR models
text-to-speech deep-learning pytorch tts speech-synthesis non-autoregressive voicebox flow-matching f5-tts mb-istft
-
Updated
Feb 23, 2026 - Python