Automatic Speech Recognition on FLEURS (test)
0.1745Average Error RateLatentASR
Evaluation results
| Method | Links | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| LatentASR2026.07 | 0.1745 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 19 | |
| Baseline2026.07 | 0.1748 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MMSModel=MMS2026.06 | 0.32 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.3 | -5.4 | — | — | — | |
| Whisper-largeModel=Whisper-large2026.06 | 0.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.64 | -8 | — | — | — | |
| Whisper-smallModel=Whisper-small2026.06 | 1.27 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 1.21 | -4.7 | — | — | — | |
| Qwen3.5-Omni-Plus2026.04 | 6.6 | 2.9 | 3.2 | 2.2 | 11.7 | 2 | 2.6 | 2.2 | 2.1 | 1.6 | 0.8 | 1.7 | 3.1 | 2.8 | 1.9 | 1.9 | 3.1 | 9.7 | 2.7 | 2.8 | 20.8 | 3.9 | 3.1 | 3.5 | 12.5 | 2.4 | 1.9 | 3.6 | 2.6 | 5.1 | 12 | 4.7 | 10.6 | 15.8 | 6.7 | 6.2 | 16.2 | 5.4 | 2.8 | 10.5 | 6.7 | 5 | 13.9 | 5.4 | 4.9 | 11.8 | 6.3 | 16 | 10 | 6.7 | 4.1 | 18.8 | 16.3 | 13.7 | 3.2 | 3.3 | 6.1 | 9.4 | 10 | 7.2 | 3.2 | — | — | — | — | — | — | — | — | |
| Gemini-1.5-Pro2026.04 | 7.3 | 3.6 | 2.7 | 6.3 | 9.2 | 2.7 | 3.7 | 2.5 | 2.6 | 2.5 | 1.1 | 2 | 3.4 | 4.3 | 2.5 | 2.3 | 3.8 | 4.5 | 3.7 | 3.5 | 25.2 | 5 | 4.6 | 5.7 | 15.6 | 3.4 | 2.7 | 4.7 | 3.8 | 7.6 | 8.9 | 5.4 | 12.7 | 23.7 | 6.7 | 5.3 | 21.9 | 6 | 2.7 | 13 | 4.9 | 4.9 | 14.8 | 5.1 | 5.5 | 14.1 | 6.2 | 16.3 | 8.3 | 3.7 | 4 | 18.3 | 15.3 | 14.4 | 3.4 | 2.8 | 6.3 | 9.9 | 20.4 | 5.8 | 3.4 | — | — | — | — | — | — | — | — | |
| Proposed (MoE + modified CIF)Training data duration (h)=80002026.06 | 8.65 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Whisper-large-v32026.06 | 9.59 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o-Transcribe2026.04 | 10.4 | 2.6 | 3.2 | 5.2 | 13 | 2.3 | 3.7 | 2.3 | 2.3 | 3.5 | 1.4 | 2.1 | 3.7 | 4.9 | 3.5 | 3 | 4.2 | 12 | 4.1 | 3.7 | 19.7 | 5.5 | 5.2 | 6.5 | 19.4 | 3.8 | 2.8 | 10.8 | 4.7 | 7.3 | 9.9 | 6.5 | 17.9 | 23.8 | 10.2 | 7 | 24.1 | 13.9 | 2.7 | 15.1 | 7.6 | 6.8 | 26.9 | 16.4 | 7.4 | 24.7 | 11.5 | 28.1 | 20.7 | 6.3 | 6.2 | 33.5 | 26.3 | 36.4 | 4.5 | 3.6 | 8.8 | 16.3 | 20.2 | 10.6 | 4.3 | — | — | — | — | — | — | — | — | |
| Proposed (MoE + modified CIF)Training data duration (h)=15002026.06 | 10.46 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini-1.5-Flash2026.04 | 10.5 | 4.6 | 2.9 | 10.8 | 10.1 | 3.3 | 3.9 | 2.7 | 2.9 | 2.8 | 1.8 | 2.4 | 3.9 | 4.5 | 3.5 | 3.4 | 4.4 | 5.6 | 6.2 | 4.7 | 23 | 6.7 | 7.7 | 7.9 | 20.2 | 5.2 | 4.9 | 6.8 | 8 | 8.5 | 10 | 7.6 | 18.6 | 48.3 | 10.9 | 7.9 | 21.9 | 11.2 | 6.4 | 12.8 | 7.3 | 14.6 | 16.3 | 9.3 | 11.3 | 15.7 | 12.7 | 16.3 | 16.3 | 6.8 | 7.8 | 20.3 | 16 | 17.4 | 5.9 | 6.5 | 10 | 10.7 | 53.1 | 13.2 | 5.1 | — | — | — | — | — | — | — | — | |
| Qwen3.5-Omni-Flash2026.04 | 10.8 | 2.9 | 3.7 | 3.1 | 13.6 | 2.5 | 3.3 | 2.4 | 2.2 | 2.4 | 1 | 2.1 | 3.6 | 3.2 | 2.5 | 2.5 | 4.4 | 9.9 | 4.2 | 3.5 | 31.9 | 5.2 | 5 | 5.3 | 16.6 | 4.5 | 3.1 | 8.9 | 4.5 | 7.1 | 12.1 | 8.1 | 13.7 | 25.9 | 12.2 | 10.7 | 19.8 | 9.5 | 6.3 | 16.6 | 16.6 | 8.6 | 18.4 | 9 | 10.6 | 18.3 | 16.6 | 23.8 | 19.7 | 17.8 | 7.9 | 27 | 23.6 | 24.6 | 6.1 | 5.5 | 14.3 | 17.5 | 41.1 | 13 | 5.4 | — | — | — | — | — | — | — | — | |
| LLM-ASR + MoE ProjectorTraining data duration (h)=15002026.06 | 11.06 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LLM-ASR + CIF DownsamplerTraining data duration (h)=15002026.06 | 12.89 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LLM-ASR BaselineTraining data duration (h)=15002026.06 | 13.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LatentASR2026.07 | 32.57 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0.23 | |
| Baseline2026.07 | 32.65 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SBPN-LargeModel size=600 M2026.05 | 32.72 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 24.38 | 33.86 | 39.94 | — | — | — | — | — | |
| SBPN-BaseModel size=120 M2026.05 | 36.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 27.04 | 39.53 | 43.83 | — | — | — | — | — | |
| MMS-1BModel size=1 B2026.05 | 41.23 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 25.51 | 44.61 | 53.56 | — | — | — | — | — | |
| MLESetup=Multilingual (ConMamba)2026.07 | 41.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.03 | 36.59 | — | |
| MLFESetup=Multilingual (ConMamba)2026.07 | 42.56 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.95 | 37.16 | — | |
| MISetup=Multilingual (ConMamba)2026.07 | 42.89 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 48.29 | 37.49 | — | |
| M-CTC+LIDSetup=Multilingual (ConMamba)2026.07 | 44.66 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 49.98 | 39.35 | — | |
| MIFSetup=Multilingual (ConMamba)2026.07 | 45.51 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 51.14 | 39.88 | — | |
| ConMambaSetup=Monolingual2026.07 | 50.95 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.26 | 45.64 | — | |
| ConformerSetup=Monolingual2026.07 | 51.27 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.57 | 46.97 | — | |
| SeamlessM4T v2Model size=2.3 B2026.05 | 90.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 96.9 | 83.5 | — | — | — | — | — | |
| Whisper LargeModel size=1.5 B2026.05 | 116.46 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 144.33 | 101.49 | 103.56 | — | — | — | — | — |