Correlation of Formant Centralization Ratio with Perceptual Measures of Dysarthria Speech

Abstract

Poor speech intelligibility is one of the main concerns in dysarthria speech. Speech intelligibility can be measured using different methods. Formant centralization ratio (FCR) is one of the quantitative methods of finding speech intelligibility based on acoustical measures. The present study aims to find if there is any correlation between FCR and perceptual measures of intelligibility. There were six subjects including four males and two females as participants for the study. All of them were diagnosed to have dysarthria secondary to cerebrovascular accident. Phonation of /a, i, u/ were taken along with narration as speech samples. For perceptual measures, three trained speech-language pathologists rated intelligibility of narration sample based on the open set method. First three formants of each vowel were measured using Praat software (Version) to find FCR value. It was calculated using the formula given by Sapir, Ramig, Spielman and Fox (2010). Cronbach’s alpha of inter and intra judge reliability was found to be more than 0.75 for the narrative sample for all the subjects. FCR was greater in severe dysarthria patients; besides, it was less in mild severity. Based on the statistical analysis, there was a positive moderate correlation between FCR and perceptual measures of speech intelligibility in dysarthric speech. And it was not gender or task-specific. The present study indicated that FCR is a valid quantitative measure of speech intelligibility along with perceptual measures. There is a need for reassuring the use of FCR by finding the extent of correlation with other acoustical measures and even with physiological methods in future.

Keywords: dysarthria, formant centralization ration, perceptual measure, phonation, speech intelligibility

Cite this Article

Irfana M., Dollin Johnson, Himani Bansal. Correlation of Formant Centralization Ratio with Perceptual Measures of Dysarthria Speech. Research & Reviews: A Journal of Bioinformatics. 2019; 6(3): 12–15p.