FunAudioLLM/Fun-ASR-Nano

Fun-ASR-Nano: LLM-Powered Speech Recognition

End-to-end ASR model trained on tens of millions of hours, supporting 31 languages including Chinese dialects (Cantonese, Sichuan, Shanghai, Minnan, etc.).

Key Features

• —🌍 31 languages + Chinese dialect recognition
• —🎯 Native punctuation output (no post-processing needed)
• —⚡ Timestamps and speaker diarization support
• —🔥 Trained on massive multilingual data

Links

• —GitHub: Fun-ASR | FunASR Toolkit
• —Model: Fun-ASR-Nano-2512