Acoustic voice analysis of Cantonese-speaking children : a comparison between MDVP and Praat

Abstract

The aim of the present study was to compare two voice analysis programs, MDVP and Praat, in analyzing voices of Cantonese-speaking children. Voice samples from a total of 42 children (21 dysphonic and 21 vocally healthy) aged 6;02 to 12;09 were obtained from the database at the Voice Research Laboratory, The University of Hong Kong. Voice samples included a sustained vowel /a/ prolongation and a Cantonese sentence /ba1 ba1 da2 gɔ1 gɔ1/. Four acoustic measures including mean fundamental frequency, jitter local, shimmer local and noise-to-harmonic ratio of the voice samples were obtained from the two programs. The values obtained from the two acoustic analysis programs and their performance in distinguishing dysphonic and normal voices were compared. The results found that values of mean fundamental frequency were similar between MDVP and Praat but values of jitter local, shimmer local and noise-to-harmonic ratio were significantly different between the two programs. Moreover, the two voice analysis programs showed similar discriminative capabilities in distinguishing dysphonic and normal pediatric voices using sustained vowel prolongation. However, when using sentence stimuli, MDVP was slightly more effective than Praat in distinguishing dysphonic and normal pediatric voices.