List your company

Audio-Visual Speech Recognition on LRS-3 Average Babble of noisy SNR levels (-10 to 5dB) (test)

9.9WERUADF
9.2213.8118.422.99Feb 8, 2024

Evaluation results

MethodLinks
UADFFusion Strategy=Late fusion2024.029.922.7
MSRLFusion Strategy=Late fusion, Additional Training Procedure=true2024.0210.121.1
Static wFusion Strategy=Late fusion2024.0210.617.2
AV-HuBERTModality=audio-visual2024.0212.8—
V-HuBERTModality=visual-only2024.0226.9—