An investigation of audio-visual speech recognition as applied to multimedia speech therapy applications

A multimedia speech therapy system should be able to be used for customized speech therapy for different problems and for different ages. The speech recognition must be designed to work with high inter- and intra-speaker variability. In addition to displaying text on a screen, recording the voice reading the text, analyzing the recorded spoken signal and performing speech recognition which includes identification of speech irregularities and tracking of patient progress, it should be capable of analyzing visual signal of the patients' speech and provide visual as well as audio feedback. This implies that the synchronization of different media is important in realizing effective multimedia speech therapy applications. In order to perform speech recognition and identification tasks, time-frequency analysis and neural networks are proposed with integration of visual information.