Voice quality as a pitch-range indicator

Pitch perception plays a central role in processing speech prosody. Since f0 varies from speaker to speaker and from context to context, effective pitch-range normalization is thus important to uncover intended linguistic pitch targets. It has also been speculated that voice quality may play a role in pitch-range perception. Our previous study demonstrated that spectral balance indeed effectively affected the perception of pitch height: “tense voice”, implemented as stimuli with spectral balance tilted towards higher frequency, was perceived as higher in pitch. Our previous study used nonspeech stimuli, raising the possibility that listeners might not be in the speech mode; this current study therefore replicates the previous experiment using speech stimuli resynthesized with the same range of f0 contours and a similar spectral manipulation, and the same forced-choice pitch classification experiment with four spectral conditions. The results are consistent with our previous experiment: the pitch classification function was significantly shifted by differences in spectral balance. Listeners generally hear higher pitches when the spectrum includes more high-frequency energy (i.e., tenser phonation). Moreover, there is a salient perceptual bias: When the second peak is tenser, the effect is stronger. These new results further support the hypothesis that voice quality cues are strong indicators of pitch-range.