Complex pitch perception with auditory model-based stimuli in humans

Pitch can be evoked by a single sinusoidal component, a harmonic complex tone, or even from a mixture of multiple harmonic complex tones. An accurate model of pitch perception should account for human perception of all these stimuli. This line of inquiry includes research by Hartmann and colleagues testing pitch models for mistuned harmonics [Lin and Hartmann, J. Acoust. Soc. Am. 103, 2608–2617 (1998)]. Building on this body of research, our study tests pitch models for multiple simultaneous pitches using “rate-place metamers,” synthetic stimuli with rate-place representations matching those of pitch-evoking stimuli but altered temporal representations. We measured F0 difference limens (F0DLs) in 20 normal-hearing human listeners for original stimuli (ORIG) and their rate-place metamers (META), across a range of F0s and spectral regions, for single harmonic complexes and triads. We also measured perception of all three pitches in a triad, with an objective major-vs.-minor task and a subjective expectation rating task. We observed similar patterns of behavioral results for ORIG and META across all tasks, broadly supporting a rate-place view of multiple pitch perception. However, some observed differences between ORIG and META cannot be well explained by the rate-place model. [Work supported by NIH, Grant R00DC017472 (AHM)]