Automatic Speech Recognition on VoxPopuli English
5.88WERCanary-1b-v2
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Canary-1b-v2Decoding Architecture=AR, Params (B)=1.02025.08 | 5.88 | 47.87 | 6.4 | |
| Qwen3-ASR-1.7BDecoding Architecture=AR, Params (B)=2.02025.08 | 5.97 | 20.12 | 5.62 | |
| Whisfusion + rerankDecoding Architecture=NAR, Params (B)=2.42025.08 | 6.34 | 143.27 | 6.07 | |
| WhisfusionDecoding Architecture=NAR, Params (B)=1.52025.08 | 6.66 | 173.45 | 6.55 | |
| Drax (MBR, 8/16)Decoding Architecture=NAR, Params (B)=1.22025.08 | 7.07 | 33.16 | 8.34 | |
| OWSM-CTC v3.1Decoding Architecture=NAR, Params (B)=1.02025.08 | 8.13 | 665.74 | 8.27 | |
| MMS-allDecoding Architecture=NAR, Params (B)=1.02025.08 | 8.53 | 681.21 | 12.93 | |
| Whisper-large-v3Decoding Architecture=AR, Params (B)=1.62025.08 | 8.96 | 35.33 | 7.11 | |
| Whisper-LLaDADecoding Architecture=NAR, Params (B)=8.72025.08 | 10.47 | 7.92 | 9.82 | |
| Whisper-turboDecoding Architecture=AR, Params (B)=0.82025.08 | 11.24 | 143.99 | 8.86 |