🎛️ OmniVoice Word-Control — voice cloning with word-level prosody control

Zero-shot voice cloning with explicit, decoupled word-level control over five acoustic dimensions, à la WordVoice — but built on OmniVoice (masked-diffusion LM TTS), fine-tuned on WordVoice-5A with inline control tokens.

Upload a reference clip (+ optional transcript — auto-transcribed if empty), write your text, and attach inline tags to any word to steer its prosody. Untagged words are planned freely.

Model · fine-tune of k2-fsa/OmniVoice