Identification of Pathological Voices Using Glottal Noise Measures

Abstract

We investigated the abilities of four fundamental frequency (F 0 )-dependent and two F 0 -independent measures to quantify vocal noise. Two of the F 0 -dependent measures were computed in the time domain, and two were computed using spectral information from the vowel. The F 0 -independent measures were based on the linear prediction (LP) modeling of vowel samples. Tests using a database of sustained vowel samples, collected from 53 normal and 175 pathological talkers, showed that measures based on the LP model were much superior to the other measures. A classification rate of 96.5% was achieved by a parameter that quantifies the spectral flatness of the unmodeled component of the vowel sample.