Deep Learning-Based Automated Lip-Reading: A Survey

A survey on automated lip-reading approaches is presented in this paper with the main focus being deep learning related methodologies which have proven to be more fruitful for both feature extraction and classification. This also provides comparisons of all different components that make up systems including audio-visual databases, extraction, classification networks schemas. The contributions unique insights are: 1) comparison Convolutional Neural Networks other neural network architectures extraction; 2) critical review advantages Attention-Transformers Temporal Recurrent classification; 3) schemas used ASCII characters, phonemes visemes, 4) most up-to-date until early 2021.