A microcomputer based system for acoustic analysis of voice characteristics

If we want to make research about voice characteristics, the effect of voice therapy and voice training or the effect of treatment of pathological voices, we need objective and simple to use methods to describe different aspects of the voice. This paper reports on an effort to include several types of acoustic voice analysis methods in one unit consisting of a microcomputer and some external hardware. Different analysis methods are described and discussed with special emphasis on automatic inverse filtering of normal and pathological voices are reported. In order to evaluate the acoustic measures, perceptual analysis of discernible voice qualities have been compared with the acoustic results. This report is the result of a joint project between the Department of Speech Communication and Music Acoustics, KTH, and the Department of Logopedics and Phoniatrics at Huddinge University Hospital.