Pitch and pitch strength of iterated rippled noise

Abstract

A cascade of add, delay (d ms), and attenuate (−1≤g≤1) circuit excited with noise produces iterated rippled noise (IRN) stimuli. The matched pitch and discriminability between pairs of IRN stimuli were studied as a function of g, d, and the number of circuit iterations (n). For g≳0, the pitch of all IRN stimuli equals 1/d. For g<0, pitch depends on n: For small n, there were two pitches in the region of 1/d, while for large n there was a single pitch equal to 1/2d. Peaks in the autocorrelation function of IRN stimuli accounted for all of the results. Peaks in the autocorrelation functions for IRN stimuli indicate the number of intervals in the waveform with durations pd (p=1,2,...,n), and for g<0 intervals related to peaks near 1/md (m=odd integers) caused by assumed auditory filtering. The number of intervals (i.e., the heights of the autocorrelation peaks) determines the discriminability between IRN stimuli, while the reciprocal of the interval duration determines the matched pitch. These results support a temporal rather than a spectral account of the pitch of IRN stimuli. [Work supported by NIH.]

Pitch strength of noise-vocoded harmonic tone complexes in normal-hearing listeners

“…The similarities in pitch strength judgments for noise-vocoded wHTCs and IIRNs establish the saliencies evoked by noise-vocoded wHTCs as meaningful in terms of pitch percepts. Previous studies on pitch strength using magnitude judgment have concluded that saliency is related to temporal properties of the stimuli (Yost, 1996b;Shofner and Selas, 2002). However, in the present study, the predicted judgments based on temporal structure (i.e., PS or AC1) do not account for the actual judgments of listeners' pitch strength [Figs.…”

Pitch strength and Stevens’s power law

“…It has been argued that pitch strength is directly related to the height of AC1 in the stimulus autocorrelation function Yost, 1996b;Yost, Patterson, & Sheft, 1996). With the use of a scaling procedure to measure the pitch strengths of iterated rippled noises, Yost (1996b) described the relationship between AC1 and pitch strength with the exponential function described by Equation 2.…”

“…If the second stimulus had a pitch strength between that of the first and third stimuli, the pitch strength of the second stimulus should be assigned a rating that is scaled relative to the pitch strengths of the first and third stimuli. This procedure is identical to that used by Yost (1996b). If needed, the listeners could listen to any trial repeatedly until they felt confident enough to assign a rating to the perceived pitch strength.…”

“…Indeed, unlike loudness, which is clearly related to the overall sound pressure level, the stimulus dimension related to pitch strength is more equivocal. More recently, however, it has been argued that pitch strength can be predicted on the basis of the height of the first peak (AC1) in the stimulus autocorrelation function (Patterson, Handel, Yost, & Datta, 1996;Yost, 1996b;Yost, Patterson, & Sheft, 1996). Yost (1996b) scaled the pitch strengths of iterated rippled noises and described the relationship between AC1 and the perceived pitch strength with an exponential function having the form y 5 a 1 (10 bAC1 ),…”

“…In neither of the studies above was pitch strength evaluated in terms of Stevens's power law. Based on the shape of the exponential function described by Yost (1996b), it would be predicted that b in Stevens's power law would be greater than 1 if AC1 were equivalent to the stimulus dimension f. Thus, if b is greater than 1, pitch strength will increase at a faster rate of change than will the height of AC1. Moreover, if pitch strength is related to the height of AC1, as described by Stevens's power law, it would be predicted that there should be no effect of relative starting phase on the pitch strengths for harmonic tone complexes.…”