A multivariate model incorporating subharmonic measurements for evaluating vocal roughness.

The assessment of voice quality plays a critical role in the clinical evaluation of hoarseness. However, no study has established a highly accurate method for the auditory-perceptual judgment of vocal roughness, which is one of the major components of hoarseness. In this study, we developed a multivariate acoustic model for quantifying vocal roughness using a tailored fundamental frequency (f) estimation algorithm. The newly devised parameters enabled the classification and quantification of subharmonics, a key component of rough voice. Furthermore, we introduce the acoustic roughness index (ARI), a predictive acoustic model that integrates these parameters with existing acoustic parameters. The ARI demonstrates high diagnostic accuracy and a strong correlation with auditory-perceptual roughness, establishing it as a robust index for the evaluation of vocal roughness.