CosyVoice3 (FunAudioLLM)

Multi-lingual large voice generation model, providing inference, training and deployment full-stack ability.

Key Features

• 0.5B lightweight model (Fun-CosyVoice3-0.5B)
• 3-second极速声音克隆
• Natural language emotion control
• Streaming support
• Fully surpasses previous generation while remaining lightweight

Open Source

• GitHub: FunAudioLLM/CosyVoice
• Model weights available on ModelScope