Voice quality distributions in a perceptual space
Vishwas Shetty, Abeer A. Alwan, Jody E. Kreiman · The Journal of the Acoustical Society of America · 2025
Recent research [Lee and Kreiman, J. Acoust. Soc. Am. 151, 3462 (2022)] suggests that important psychoacoustic space. However, without knowledge of how voices are distributed in this space, it is difficult to speculate about how listeners exploit its structure perceptually, or to predict which voices sound similar or different. Building on prior studies employing principal component analysis, we defined a voice space using the first two PCs that emerged from analyses of acoustic variability in voice, and examined how voices vary in their distributions within that space. The center of the distributions for individual voices in this space varied minimally across speakers, but other features of the distributions, such as the nature and extent of variability within the space, revealed substantial individual differences. These nuances of vocal variability provide insights into the psychoacoustic structure of voices, highlighting how individual patterns of variability are distributed across a shared voice space. Our analysis also explores how such variability impacts speaker verification, with implications for the accuracy and robustness of automatic systems.