Automatic Speech Recognition on AISHELL-1
0.6WERKimi Audio
Evaluation results
| Method | Links | |
|---|---|---|
| Kimi Audio2025.10 | 0.6 | |
| Kimi-AudioSize=7B2025.09 | 0.61 | |
| Kimi-Audio-7BReference audio=None2026.04 | 0.61 | |
| UAF-30B-A3BReference audio=None2026.04 | 0.84 | |
| Qwen3-Omni-30B-A3BReference audio=None2026.04 | 1.03 | |
| Qwen3 Omni2025.10 | 1.04 | |
| Ming-Flash-Omni2025.10 | 1.11 | |
| Qwen2.5-Omni-7BReference audio=None2026.04 | 1.13 | |
| Qwen2-Audio-7BReference audio=None2026.04 | 1.52 | |
| Qwen2 Audio2025.10 | 1.53 | |
| Covo-Audio-BaseParams=7B2026.02 | 1.96 | |
| GLM-4-Voice-BaseParams=9B2026.02 | 2.46 | |
| Paraformer-zh-streamingReference audio=None2026.04 | 3.05 | |
| VITA-AudioSize=7B2025.09 | 4.58 | |
| Pretrain+TtTSize=3B2025.09 | 5.78 | |
| Whisper-large-v3Params=1.5B2026.02 | 9.31 | |
| TtTSize=3B2025.09 | 13.65 | |
| ERNIE 5.02026.02 | 31 | |
| Kimi Audio2026.02 | 60 | |
| Kimi-AudioSize=9B2026.04 | 60 | |
| Qwen3-OmniSize=30B-A3B2026.04 | 60 | |
| LongCat-Flash-Omni2026.02 | 63 | |
| Qwen2.5-3B (AR)Size=3B2025.09 | 72.01 | |
| ERNIE 5.0-Base2026.02 | 75 | |
| Qwen3-Omni-Instruct2026.02 | 84 | |
| MiniCPM-o 4.5Size=9B2026.04 | 90 | |
| SpeechGPTSize=7B2025.09 | 111.81 | |
| Qwen2.5-3B (NAR)Size=3B2025.09 | 160.58 | |
| Gemini-3-Pro2026.02 | 304 | |
| GPT-4o-Audio2026.02 | 352 | |
| Mini-OmniSize=0.5B2025.09 | 442.06 |