hugpadhugpad

k2-fsa / OmniVoice

Audio613M3M1.6KUpdated Jul 3, 2026
Hugging Face

Downloads

3M

Likes

1.6K

Last updated

Jul 3, 2026

License

Not available

OmniVoice is a massively multilingual zero-shot text-to-speech (TTS) model supporting over 600 languages. Built on a novel diffusion language model-style architecture, it delivers high-quality speech with superior inference speed, supporting voice cloning and voice design.

Tasks & library

  • text to speech
  • omnivoice

Tags

  • omnivoice
  • zero-shot
  • multilingual
  • voice-cloning
  • voice-design
  • text-to-speech

Technical specifications

Parameters613M
Libraryomnivoice
Languages646 languages
Base modelQwen/Qwen3-0.6B
LicenseNot available