Voice Quality Assessment and Visualization

Abstract

Commercial voice analysis systems are expensive and complicated. These equipments are only available at medical centers and need to be operated by well trained physicians. Thus the treatment costs of speech disorders are usually high. To improve this situation, we develop a software system dedicated to the assessment of human voice qualities. Our system is implemented in an ordinary personal computer. High precision electronic devices are not required for acquiring and analyzing voices. Therefore, the installation costs are low. Our system relies on four voice parameters to evaluate the qualities of voice signals. These parameters are fundamental frequency, jitter, shimmer, and harmonic-to-noise ratio. These measurements are widely used in otolaryngology to assess the pitch, variation of frequency, perturbation of amplitude, and harmony of human voice. Our system extracts these information from voice data and displays them by using graphical media. Consequently, the qualities of voice are more comprehensible. Users with little training and background knowledge can operate this system in their living rooms to assess their voices.