Psychometric features of Acoustic Voice Quality Index, version 03.01, in Persian language

Abstract

ABSTRACT The goal of this study was to investigate the psychometric features of the Acoustic Voice Quality Index, in Persian, as an objective and multi-parameter tool for evaluating the overall voice quality. Three hundred and ten Persian-speaking individuals, 155 with various types and severity of dysphonia, and 155 healthy individuals, performed two tasks: reading a phonetically balanced text and a sustained vowel /a/. Voice recordings were made. Perceptual evaluation with the G score of the GRBAS scale was completed by five raters, and intra-rater reliability and inter-rater reliability were calculated. The Acoustic Voice Quality Index was calculated with standard syllables corresponding to 3 s of voiced speech and 3 s of vowels. Concurrent validity, construct validity, test–retest reliability, and diagnostic accuracy of the AVQI were analyzed. The results suggest that the number of syllables corresponding to 3 s of voiced speech was 23. Perceptual evaluation demonstrated excellent intra-rater and inter-rater reliability. The AVQI demonstrated a strong correlation with perceptual evaluation (rs = 0.877, P < 0.001), as well as high construct validity. In addition, the index had very good test–retest reliability (icc = 0.992). The cut-off point of the index was 3.62 in Persian. The area under the ROC curve was 0.815, indicating that the index had the ability to differentiate between dysphonic and non-dysphonic voices. And it can be stated that the AVQI 03.01 is a reliable objective tool for distinguishing dysphonic from non-dysphonic voice in Persian that can be utilized in clinical and research settings.