🎀 OmniVoice Singing Demo

A finetune of OmniVoice that adds singing and emotional speech on top of state-of-the-art text-to-speech for 600+ languages:

  • Singing & Emotion β€” Prefix your text with [singing] or an emotion tag ([happy], [sad], [angry], [excited], [calm], [nervous], [whisper])
  • Voice Clone β€” Clone any voice from a reference audio
  • Voice Design β€” Create custom voices with speaker attributes

Model: ModelsLab/omnivoice-singing Β· Built with OmniVoice by Xiaomi AI Lab Next-gen Kaldi team.

Prefix your text with a tag. Combine them too, e.g. [singing] [sad] .... A Guidance Scale of 3.0 is recommended for pronounced tag behavior.

Language (optional) / 语种 (可选)

Keep as Auto to auto-detect the language.

Examples