Automatic Speech Recognition on LRS3 433-hour labeled (test)
1.3WER (%)AV-HuBERT
Evaluation results
| Method | Links | |
|---|---|---|
| AV-HuBERTEncoder Size=325M, Criterion=CE, Setting=Large Block 22025.02 | 1.3 | |
| RAVenEncoder Size=671M, Criterion=CTC+CE2025.02 | 1.4 | |
| u-HuBERTLabel Data=433h, Encoder Size=325M, Criterion=CE2025.02 | 1.4 | |
| AV-data2vecEncoder Size=325M, Criterion=CE, Setting=Large Block 22025.02 | 1.4 | |
| DistillAVEncoder Size=325M, Criterion=CE+KD, Setting=Large Block 22025.02 | 1.4 | |
| AV-data2vecEncoder Size=103M, Criterion=CE, Setting=Base Block 22025.02 | 1.7 | |
| DistillAVEncoder Size=325M, Criterion=CE+KD2025.02 | 1.8 | |
| DistillAVEncoder Size=103M, Criterion=CE+KD2025.02 | 1.9 | |
| RAVenEncoder Size=97M, Criterion=CTC+CE, Setting=Base Block 22025.02 | 1.9 | |
| DistillAVEncoder Size=103M, Criterion=CE+KD, Setting=Base Block 22025.02 | 1.9 | |
| AV-data2vecUnlabeled Data=433h, Label Data=433h, Encoder Size=325M, Criterion=CE2025.02 | 1.9 | |
| AV-data2vecUnlabeled Data=433h, Label Data=433h, Encoder Size=103M, Criterion=CE2025.02 | 2 | |
| AV-HuBERTEncoder Size=103M, Criterion=CE, Setting=Base Block 22025.02 | 2 | |
| RAVenEncoder Size=97M, Criterion=CTC+CE2025.02 | 2.2 | |
| AV2vec-MLMEncoder Size=103M, Criterion=CE2025.02 | 2.7 | |
| AV-HuBERTEncoder Size=325M, Criterion=CE2025.02 | 2.7 | |
| AV-HuBERTEncoder Size=103M, Criterion=CE2025.02 | 3 | |
| Makino et al.Label Data=31kh, Criterion=Transducer2025.02 | 4.8 | |
| Afouras et al.Label Data=1519h, Criterion=CE2025.02 | 8.3 |