Audio-Visual Speech Recognition on LRS2 Random noise (CER by SNR)
16.52CER (-10 dB)VisG AV-HuBERT
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| VisG AV-HuBERTArchitecture=Large2026.04 | 16.52 | 12.4 | 10.76 | 10.15 | 9.71 | |
| AV-HuBERTArchitecture=Large2026.04 | 17.23 | 13.05 | 10.97 | 10.72 | 10.54 | |
| AV-HuBERTArchitecture=Base2026.04 | 19.52 | 14.64 | 11.28 | 10.03 | 9.46 | |
| VisG AV-HuBERTArchitecture=Base2026.04 | 20.34 | 14.74 | 11.43 | 10.27 | 9.47 |