A Comprehensive Application for Grading Severity of Voice Based on Acoustic Voice Quality Index v.02.03

Srushti Shabnam; M Pushpavathi; R Gopi Sankar; Karthik Venkat Sridharan; M S Vasanthalakshmi

doi:10.1016/j.jvoice.2022.08.013

A Comprehensive Application for Grading Severity of Voice Based on Acoustic Voice Quality Index v.02.03

J Voice. 2022 Sep 30:S0892-1997(22)00245-4. doi: 10.1016/j.jvoice.2022.08.013. Online ahead of print.

Authors

Srushti Shabnam¹, M Pushpavathi², R Gopi Sankar³, Karthik Venkat Sridharan⁴, M S Vasanthalakshmi⁵

Affiliations

¹ Nitte Institute of Speech and Hearing, Mangalore, Karnataka, India. Electronic address: simpleshabnam@gmail.com.
² All India Institute of Speech and Hearing, Mysuru, Karnataka, India.
³ Department of Clinical Services, All India Institute of Speech and Hearing, Mysuru, Karnataka, India.
⁴ Department of Electronics, All India Institute of Speech and Hearing, Mysuru, Karnataka, India.
⁵ Department of Speech Language Pathology, All India Institute of Speech and Hearing, Mysuru, Karnataka, India.

PMID: 36192290
DOI: 10.1016/j.jvoice.2022.08.013

Abstract

Acoustic Voice Quality Index is a six-variable acoustic model for the multiparametric measurement developed by Maryn et al. Studies have provided evidence regarding the practical usefullness, internal consistency, external validity, diagnostic accuracy, and responsiveness to change of AVQI. Recently, researchers have been exploring the utility of AVQI in classifying the voice severity. The aim of the present study was to determine the diagnostic accuracy of the AVQI v.02.03 in discriminating across the perceptual levels of dysphonia severity in 18-40 years age range in Kannada speaking population; and to develop an application to depict the AVQI based severity of dysphonia. For the study, 163 individuals in normophonic and 134 individuals in dysphonic group were considered in the age range of 18-40 years. All participants were native speakers of Kannada language. The sustained vowel /a/ and reading of standard Kannada passage were considered as stimuli for extracting AVQI analysed using AVQI script version 02.03. The AVQI cut-off values obtained were 2.50 (A_ROC=0.894; Sensitivity= 84.7%; Specificity= 83.1%), 4.17 (A_ROC=0.953; Sensitivity= 84.4%; Specificity= 88.5%) and 6.23 (A_ROC=1.000; Sensitivity= 100%; Specificity= 100%) for normal vs. mild, mild vs. moderate and moderate vs. severe respectively. A user friendly application was developed which provides a simplified output for AVQI cut-off values which can be comprehendible by patients with voice disorder/ non-professionals and health professionals.

Keywords: AVQI v.02.03; Acoustic; Acoustic voice quality index; Analysis; Dysphonia; Voice disorders.