Transformer-Based Video Front-Ends for Audio-Visual Speech Recognition for Single and Multi-Person Video
This work improves speech recognition accuracy for applications like video transcription by integrating visual cues, though it is incremental as it adapts an existing transformer method to a new modality.
The paper tackled audio-visual speech recognition by using a video transformer to extract visual features from speaker mouth motion, achieving state-of-the-art performance with 1.6% WER on LRS3-TED and relative improvements of 10-15% over convolutional baselines on video-only tasks.
Audio-visual automatic speech recognition (AV-ASR) extends speech recognition by introducing the video modality as an additional source of information. In this work, the information contained in the motion of the speaker's mouth is used to augment the audio features. The video modality is traditionally processed with a 3D convolutional neural network (e.g. 3D version of VGG). Recently, image transformer networks arXiv:2010.11929 demonstrated the ability to extract rich visual features for image classification tasks. Here, we propose to replace the 3D convolution with a video transformer to extract visual features. We train our baselines and the proposed model on a large scale corpus of YouTube videos. The performance of our approach is evaluated on a labeled subset of YouTube videos as well as on the LRS3-TED public corpus. Our best video-only model obtains 31.4% WER on YTDEV18 and 17.0% on LRS3-TED, a 10% and 15% relative improvements over our convolutional baseline. We achieve the state of the art performance of the audio-visual recognition on the LRS3-TED after fine-tuning our model (1.6% WER). In addition, in a series of experiments on multi-person AV-ASR, we obtained an average relative reduction of 2% over our convolutional video frontend.