While widely used speech recognition tools like Siri or Otter generally analyze audio alone, researchers have also made progress in developing visual speech recognition (VSR) models, which rely on visual input to identify what a speaker is saying.