🚀 Fun-ASR-Nano
LLM-Powered Speech Recognition — 31 Languages, Dialects & Accents
End-to-end ASR trained on tens of millions of hours of data. Supports Chinese (+ dialects), English, Japanese, Korean, French, German, Spanish, and 24 more languages.
⭐ GitHub (Fun-ASR) · 🛠️ FunASR Toolkit · 🎙️ SenseVoice · 🤗 Model Card
| Model | Languages | Architecture | Best For |
|---|---|---|---|
| Fun-ASR-Nano ⭐ | 31 | LLM-based | Multi-language, dialects, highest accuracy |
| SenseVoice | 5 | CTC (non-AR) | Speed + Emotion + Audio events |
Model
Language (SenseVoice only)
Supported Languages (Fun-ASR-Nano)
Chinese (Mandarin, Cantonese, Sichuan, Shanghai, Minnan, Wenzhou, Hakka, Gan, and more), English, Japanese, Korean, French, German, Spanish, Italian, Portuguese, Russian, Arabic, Hindi, Thai, Vietnamese, Indonesian, Malay, Turkish, Polish, Dutch, Swedish, Hebrew, Greek, Czech, Romanian, Hungarian, Finnish, Danish, Norwegian, Ukrainian.
Tips
- Fun-ASR-Nano: Best for multi-language & Chinese dialects. Outputs punctuation natively.
- SenseVoice: Ultra-fast (7x faster than Whisper-small), also detects emotions & audio events.
- For long audio (>5min), consider using FunASR locally with GPU.