Automatic Speech Recognition on FLEURS (zh, en)
2.71WER (en)GPT-4o-trans.
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-4o-trans.2026.05 | 2.71 | 2.62 | |
| MEGA-ASRrouter=true2026.05 | 3.17 | 3.86 | |
| Qwen3-ASR-1.7B2026.05 | 3.19 | 3.93 | |
| MEGA-ASR2026.05 | 3.76 | 5.43 | |
| Voxtral-Mini-3B-25072026.05 | 3.84 | — | |
| Step-Audio-2-mini2026.05 | 3.93 | 2.48 | |
| Gemini-3-Flash2026.05 | 4.01 | 7.52 | |
| Qwen2.5-Omni-7B2026.05 | 4.01 | 4.64 | |
| Whisper Large-v32026.05 | 4.1 | 7.7 | |
| Canary-1B-v22026.05 | 4.48 | — | |
| Parakeet-TDT-0.6B-v32026.05 | 4.88 | — | |
| Doubao-LLM ASR2026.05 | 7.22 | 2.92 | |
| Kimi-Audio-7B2026.05 | 8.07 | 5.88 |