Combining displays of EPG and automatic segmentation of speech for clinical purposes

Abstract

The development of dynamic electropalatography (EPG) and automatic segmentation of speech has been motivated by the requirements of different end users, although both have been used for clinical purposes. EPG is used in experimental phonetics to record lingual-palatal contacts during continuous speech and there is also a growing interest in its use in speech therapy. Automatic segmentation and labelling is a technique which assigns acoustic-phonetic labels to continuous speech. It has been claimed to be useful as a front end in automatic speech recognition, and automatic language recognition. On the clinical side, coarse phonetic features have been used successfully in therapy to provide visual feedback for the deaf, such as those used in the dedicated hardware system called C Speech and also to assist in lipreading. At Leeds University, a display of automatically produced broad phonetic classes is being developed as a potential visual feedback system for the deaf. To make the system more comprehensive an...