Abstract:
Iterated rippled noise (IRN) stimuli are generated by a cascade of delay and add networks; where delay, gain after delay, and number of iterations of delay and add are the variables controlling IRN stimuli. IRN stimuli produce a sound with a tonal and a noisy percept. The tonal component is related to regularity in the stimulus and the noisy component is related to a lack of regularity. Autocorrelation is a useful measure for describing the degree of regularity in IRN stimuli. However, the envelope of IRN stimuli also has a regular and an irregular structure that can be revealed by autocorrelation. Thus, the perception of IRN stimuli could be based on fine structure and/or envelope. Autocorrelation based on fine structure is different when the delayed noise is subtracted (gain<0) than when it is added (gain>0) to the undelayed noise. However, the autocorrelation function for the envelope is the same for both addition and subtraction. Discrimination between IRN stimuli generated with addition and subtraction as a function of highpass filtering suggests that processing of IRN stimuli is based on fine structure. [Work supported by NIDCD.]