Automatic Speech Recognition on AISHELL-1 (test metric)
0.76Word Error Rate (WER)Kimi-Audio-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| Kimi-Audio-7B2026.05 | 0.76 | |
| Step-Audio-2-mini2026.05 | 0.81 | |
| Doubao-LLM ASR2026.05 | 0.98 | |
| Qwen2.5-Omni-7B2026.05 | 1.15 | |
| MEGA-ASR2026.05 | 1.49 | |
| Qwen3-ASR-1.7B2026.05 | 1.52 | |
| MEGA-ASRrouter=true2026.05 | 1.53 | |
| Gemini-3-Flash2026.05 | 2.66 | |
| GPT-4o-trans.2026.05 | 3.52 | |
| Whisper Large-v32026.05 | 5.89 |