史上最详细唇语识别最新研究进展记录

本文是唇语识别近2年来最新的方法的记录，主要集中在中英文词级数据集如LRW，LRW-1000，英文句子级数据集如LRS2，英文短语级数据集如OuluVS2，以及其他一些数据集。
记录方法为1.简要翻译重点内容；2.介绍相关文章。
记录将持续更新，对唇语识别感兴趣的研究者请收藏或私信交流，本文将持续更新。

下面是文章记录列表，链接为已记录（持续更新）：

Word-Level

1.Time Shift Module

来源：ICASSP2021，作者单位：新疆大学。
1.1.史上最详细How to Use Time Information Effectively Combining with Time Shift Module for Lipreading文章记录

2.Multi-Scale Temporal Convolutional Networks

来源：ICASSP2020，作者单位：帝国理工大学。
2.1.史上最详细 Lipreading using Temporal Convolutional Networks 环境配置
2.2.史上最详细Lipreading using Temporal Convolutional Networks(MS-TCN)代码层面详解

3.HP-ResNet18 and MS-TCN+Attention

来源：INTERSPEECH2021，作者单位：中国科学技术大学
2.1史上最详细Lip-reading with Hierarchical Pyramidal Convolution and Self-Attention文章记录

4.Synchronous Bidirectional Learning for Multilingual Lip Reading

5.Visual Keyword Spotting with Attention (also in sentence-level)

x.Effective Lip Reading

来源：ICASSP2020，作者单位：中科院计算所。

x.LiRA: Learning Visual Speech Representations from Audio through Self-supervision

x.Lip-reading with Densely Connected Temporal Convolutional Networks

x.Seeing voices and hearing voices learning discriminative embeddings using cross-modal self-supervision

x.Can We Read Speech Beyond the Lips? Rethinking RoI Selection for Deep Visual Speech Recognition

x.Mutual Information Maximization for Effective Lip Reading

x.Deformation Flow Based Two-Stream Network for Lip Reading

x.Towards Practical Lipreading with Distilled and Efficient Models

x.Multi-Modality Associative Bridging Through Memory: Speech Sound Recollected From Face Video

Sentence-Level

1.End-to-end Audio-visual Speech Recognition with Conformers

来源：ICASSP2021，作者单位：帝国理工大学。

2.Visual Keyword Spotting with Attention (also in word-level)

x.Sub-word Level Lip Reading With Visual Attention

Phrase-Level

1.Multi-Perspective LSTM for Joint Visual Representation Learning