Contribution of Fundamental Frequency to Perceived Vocal Roughness: Evidence From Time-Scaled Natural Voice Samples

Fundamental frequency (f_o) influences auditory roughness in psychoacoustic paradigms; however, it remains unclear how changes in f_o alone, when applied to dysphonic voices, alter perceived vocal roughness and breathiness. We investigated how roughness and breathiness change when f_o is systematically manipulated in dysphonic voices while maintaining the time-domain waveform shape.

Natural sustained vowels from dysphonic or nondysphonic speakers were resynthesized in Praat by time-scaling so that f_o was converted from the original value to a target f_o of 50, 100, or 200 Hz without altering the overall waveform shape (Experiment 1). In Experiment 2, an extended low-f_o set was created by converting f_o to 30-100 Hz in 10-Hz steps. Experienced raters evaluated each token for roughness and breathiness. Linear mixed-effects models tested whether changes in f_o predicted perceptual ratings after accounting for within-token and between-rater variability. Receiver operating characteristic (ROC) analysis was used to identify an f_o cutoff associated with higher roughness in Experiment 2.

Across experiments, lowering f_o systematically increased roughness ratings, whereas breathiness showed comparatively smaller and less consistent changes. In mixed-effects models, f_o was an independent predictor of roughness but not of breathiness. In Experiment 2, ROC analysis identified an f_o threshold of 70 Hz for differentiating higher versus lower roughness. These patterns are consistent with auditory accounts in which reduced harmonic spacing relative to cochlear bandwidths enhances within-band interactions and temporal-envelope modulation.

By manipulating f_o while preserving waveform shape, we identified an auditory contribution of f_o to perceived vocal roughness that is largely distinct from breathiness. These findings support the development of acoustic roughness indices that account for fo-dependent harmonic spacing and critical-band envelope modulation, and they underscore the need to consider f_o when interpreting auditory-perceptual roughness ratings.