Visual speech recognition (VSR), or automatic lip-reading, is the task of extracting speech information from visual input. The addition of visual speech has been shown to improve the performance of traditional audio ...