Visual Speech Recognition with Loosely Synchronized Feature Streams
Wepresentanapproachtodetectingandrecognizingspokenisolatedphrasesbasedsolelyonvisualinput.Weadoptanarchitecturethatrstemploysdiscriminativedetectionofvisualspeechandarticulatoryfeatures,andthenperformsrecognitionusingamodelthataccountsfortheloosesynchroniz
暂无评论