Automatic Speech Recognition on GigaSpeech
0.0845WERQwen3-ASR-1.7B
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen3-ASR-1.7BParameters=1.7B2026.01 | 0.0845 | |
| Qwen3-ASR-0.6BParameters=0.6B2026.01 | 0.0888 | |
| Gemini-2.5-Pro2026.01 | 0.0937 | |
| Qwen2.5-Omni-7B + CoATBackbone=Qwen2.5-Omni-7B, Method=CoAT2026.06 | 0.0942 | |
| Doubao-ASR2026.01 | 0.0955 | |
| Whisper-large-v32026.01 | 0.0976 | |
| Whisper-large-v3-turbo# Total Params=1.5B2025.11 | 0.1014 | |
| Audio Flamingo 3Backbone=Audio Flamingo 3, Method=Baseline2026.06 | 0.1027 | |
| Qwen2.5-Omni-7BBackbone=Qwen2.5-Omni-7B, Method=Baseline2026.06 | 0.1042 | |
| Qwen2-Audio + CoATBackbone=Qwen2-Audio, Method=CoAT2026.06 | 0.1052 | |
| LFM2-Audio-1.5B# Total Params=1.5B2025.11 | 0.1063 | |
| Qwen2.5-Omni-3B# Total Params=5B2025.11 | 0.1176 | |
| Audio Flamingo 3 + CoATBackbone=Audio Flamingo 3, Method=CoAT2026.06 | 0.119 | |
| Qwen2-AudioBackbone=Qwen2-Audio, Method=Baseline2026.06 | 0.1353 | |
| GPT-4o-Transcribe2026.01 | 0.255 |