Use of Cepstral Analyses for Differentiating Normal From Dysphonic Voices: A Comparative Study of Connected Speech Versus Sustained Vowel in European Portuguese Female Speakers

Abstract

The aim of this study was to investigate the use of cepstral peak prominence (CPP) and CPP-smoothed (CPPs) to differentiate dysphonic from nondysphonic voices, using two speech tasks: sustained vowel /a/ and connected speech. A retrospective study was based on data selected from an archival database of recorded voices. Sixty age- and occupation-matched individuals (30 participants with dysphonia and 30 controls) were recorded producing the sustained vowel /a/ and reading the European Portuguese version of "The Story of Arthur the Rat." Recorded voices were analyzed acoustically by measuring CPP and CPPs and auditory-perceptual ratings were related to the acoustic measurements. For the sustained vowel, both CPP and CPPs measures were significantly different between dysphonic and control groups. For connected speech, only CPP values revealed significant differences between the two groups, both indirect and narrative speech. Acoustic measurements correlated with the auditory-perceptual classifications in both sustained vowel and connected speech, although the strongest correlation (0.6<r<0.7) was obtained between CPP and the perception of breathiness. The results of this study suggest that analysis of CPP and CPPs is a promising tool in clinical practice with European Portuguese speakers.