List your company

Audio-Visual Speech Recognition on LRS-3 Babble noise at -5dB SNR (test)

10.7WERUADF
10.05214.42618.823.174Feb 8, 2024

Evaluation results

MethodLinks
UADFFusion Strategy=Late fusion2024.0210.720.7
MSRLFusion Strategy=Late fusion, Additional Training Procedure=true2024.0211.316.3
Static wFusion Strategy=Late fusion2024.0211.415.6
AV-HuBERTModality=audio-visual2024.0213.5—
V-HuBERTModality=visual-only2024.0226.9—