An End-to-End Mandarin Audio-Visual Speech Recognition Model with a Feature Enhancement Module | IEEE Conference Publication | IEEE Xplore