Automatic Speech Recognition on LibriSpeech 80 (test-other)
3.4WERQwen3-ASR-1.7B
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen3-ASR-1.7BModel category=ASR-Experts, Evaluation engine=Whisper-large-v32026.03 | 3.4 | |
| Qwen2.5-Omni-7BModel category=Perception-centric, Evaluation engine=Whisper-large-v32026.03 | 3.4 | |
| Qwen2-AudioModel category=ASR-Experts, Evaluation engine=Whisper-large-v32026.03 | 3.6 | |
| MinMoModel category=ASR-Experts, Evaluation engine=Whisper-large-v32026.03 | 3.9 | |
| SpeechVerseModel category=ASR-Experts, Evaluation engine=Whisper-large-v32026.03 | 4.4 | |
| HyperCLOVAX-8B-OmniModel category=Unified, Evaluation engine=Whisper-large-v3, Results reproduced=true2026.03 | 4.5 | |
| Dynin-OmniModel category=Unified, Evaluation engine=Whisper-large-v3, block size=16, diffusion steps=1282026.03 | 4.8 | |
| OpenOmniModel category=Perception-centric, Evaluation engine=Whisper-large-v32026.03 | 7 | |
| NExT-OMNIModel category=Unified, Evaluation engine=Whisper-large-v3, support for video generation=true2026.03 | 7 | |
| VITA-1.5Model category=Perception-centric, Evaluation engine=Whisper-large-v32026.03 | 7.2 |