A comparison of Dysphonia Severity Index and Acoustic Voice Quality Index measures in differentiating normal and dysphonic voices

The aim of the study was to investigate and compare the feasibility and robustness of the Acoustic Voice Quality Index (AVQI) and the Dysphonia Severity Index (DSI) in diagnostic accuracy, differentiating normal and dysphonic voices.

A group of 264 subjects with normal voices (n = 105) and with various voice disorders (n = 159) were asked to read aloud a text and to sustain the vowel /a/. Both speech tasks were concatenated, and perceptually rated for dysphonia severity by five voice clinicians. They rated the Grade (G) and the overall dysphonia severity with a visual analog scale (VAS). All concatenated voice samples were acoustically analyzed to receive an AVQI score. For DSI analysis, the required voice parameters were obtained from the sustained phonation of the vowel /a/.

The results achieved significant and marked concurrent validity between both auditory-perceptual judgment procedures and both acoustic voice measures. The DSI threshold (i.e., DSI = 3.30) pertaining to G_mean obtained reasonable sensitivity of 85.8% and specificity of 83.4%. For VAS_mean, the DSI threshold of 3.30 was determined also with reasonable sensitivity of 70.3% and excellent specificity of 93.9%. Also, the AVQI threshold (i.e., AVQI = 3.31) pertaining to G_mean demonstrated reasonable sensitivity of 78.1% and excellent specificity of 92.0%. For VAS_mean, an AVQI threshold of 3.33 was determined with excellent sensitivity of 97.0% and reasonable specificity of 81.8%.

The outcomes of the present study indicate comparable results between DSI and AVQI with a high level of validity to discriminate between normal and dysphonic voices. However, a higher level of accuracy was yielded for AVQI as a correlate of auditory perceptual judgment suggesting a reliable voice screening potential of AVQI.