Fundamental study toward standardization of acoustic evaluation of pathologic voice

点击翻译按钮获取中文摘要

Needs for establishing standard methods are increasing to evaluate pathologic voice acoustically, which is primarily motivated by a trend of evidence-based medicine at voice related clinics. Fur- thermore, the methods are expected to be realized based on web-based technology so that many users are able to access to the most contemporarily organized evaluation system. We made a com- parative study of three well-known methods to make an acoustic evaluation of pathologic voice using both natural and synthetic speech: MDVP of KayPentax which is a rather defacto standard in this area, Praat delivered by University of Amsterdam, and L-Voice developed at Utsunomiya Uni- versity. L-voice provided most reliable performance in the measurement of jitter and glottal noise, whereas Praat showed the best results for shimmer detection. A pilot system on the basis of L-Voice which is accessible over the internet has been implemented using web-technology and its efficacy has been shown in clinical assessment experiments. The clinical evaluation also indicated im- portance of GUI of the system.