Shanghai Jiao Tong / others · 2024 · flow matching
Fully non-autoregressive TTS with a Diffusion Transformer and flow matching, after E2 TTS. No duration model, no phoneme aligner — pad text to speech length and denoise. Sway Sampling at inference. MIT code; check weight licenses.
Simple recipe, strong zero-shot, a lot of local fine-tunes.
Vendor / paper demo
Zero-shot (Seed-TTS prompt) · F5-TTS official demo, Seed-TTS English zero-shot (same-language generation).
I don't really care what you call me. I've been a silent spectator, watching species evolve, empires rise and fall. But always remember, I am mighty and enduring. Respect me and I'll nurture you; ignore me and you shall face the consequences.
Vendor / paper demo
IndexTTS comparison, Speaker 1 · F5-TTS clip from the IndexTTS comparison table, Speaker 1.
Old will is a fine fellow but poor and helpless since missus rogers had her accident.