Audio-Visual Speech Recognition is Worth 32×32×8 Voxels.

Dmitriy Serdyuk Otavio Braga Olivier Siohan

Published in: CoRR (2021)

Keyphrases

audio visual speech recognition
multi stream
audio visual
pattern recognition
information retrieval
high level
hidden markov models
motion estimation
dimensionality reduction
multi modal
visual features
speech recognition
noisy environments