Audio-visual Speech Recognition on LRS3 Object occlusion and noise
25.8WER (Babble, -10 dB)
CAV2vec
Updated 3d ago
Evaluation Results
| Method | Links | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2025.12 | 25.8 | 11.7 | 4.4 | 2.4 | 1.8 | 9.2 | 5.9 | 3.6 | 2.5 | 2.1 | 1.8 | 3.2 | 9.6 | 4.3 | 2.6 | 1.8 | 1.7 | 4 | 5.1 | 7.2 | 1.5 | |
2025.12 | 30.1 | 14.3 | 5.3 | 2.5 | 1.7 | 10.8 | 6.8 | 4.4 | 2.8 | 2.3 | 2 | 3.7 | 10.5 | 5.3 | 2.8 | 2.1 | 1.8 | 4.5 | 5.9 | 8.4 | 1.6 | |
2025.12 | 30.6 | 14.4 | 5.1 | 2.7 | 2.1 | 11 | 7.7 | 4.3 | 3 | 2.2 | 2 | 3.9 | 10.9 | 5.2 | 3 | 2.3 | 1.8 | 4.6 | 6 | 8.6 | 1.6 | |
2025.12 | 32.1 | 15.1 | 5.3 | 2.5 | 2 | 11.4 | 8.3 | 4.9 | 3.1 | 2.2 | 1.9 | 4.1 | 10.9 | 5.5 | 3 | 2.1 | 1.8 | 4.6 | 6.2 | 9 | 1.5 | |
2025.12 | 41.1 | 22.2 | 6.1 | 2.5 | 1.8 | 14.7 | 45.6 | 30.4 | 14.9 | 5.2 | 2.2 | 19.7 | 20.3 | 8.3 | 3.5 | 2.1 | 1.8 | 7.2 | 12.2 | 18.7 | 1.8 | |
2025.12 | 43.5 | 25.4 | 8.5 | 4 | 2.8 | 16.9 | 58.9 | 42.3 | 17.9 | 5.1 | 3.1 | 25.5 | 23.3 | 11.3 | 5.4 | 3.3 | 2.5 | 9.2 | 15.2 | 23 | 2.3 | |
2025.12 | 54.7 | 31.6 | 14.6 | 7.3 | 5.4 | 22.7 | 45.1 | 29.3 | 16.9 | 10 | 6.6 | 21.6 | 32.1 | 17.6 | 9.1 | 6.5 | 5.2 | 14.1 | 18.1 | 25.8 | 4.2 |