Visual speech detection using OpenCV

Visual information from the human face; lip-movements and tongue provide us with lots of information about the spoken message and helps in understanding the verbal communication. The visual speech detection overcomes some of the persistent problems and inaccuracies encountered by users that creep in when there is background noise. In noisy environment we pay more attention to the lips which dramatically improves our understanding of what other people are saying. This research is focussed towards creation of a speech detector which works solely on video data. This work is part of speaker identification problem in videos. We propose speaker identification using visual clues only. Based on visual information, presence of speech can be extracted in video sequences.