ASA 127th Meeting M.I.T. 1994 June 6-10

4pPP18. Iterated rippled noise: Testing temporal theories of complex pitch.

William A. Yost

Stanley Sheft

Parmly Hear. Inst., Loyola Univ., 6525 N. Sheridan Rd., Chicago, IL 60626

Roy Patterson

MRC, Cambridge CB2 2EF, UK

Two versions of iterated rippled noise were generated by cascading a delay (T), attenuate (g), and add network n times. In one version, the new sum was delayed, attenuated, and added to the original noise iteratively. In the other version, the new sum was delayed, attenuated, and added back to itself iteratively. Pitch varies with T; and pitch strength varies with g and n, increasing as g decreases and n increases. Listeners discriminated between two iterated rippled noises based on pitch strength, and performance was measured as a function of T, n, and g. The data were accounted for by peaks in the autocorrelation functions. Listeners were often at or near chance in discriminating one noise from another when their spectra were different, but the autocorrelation functions shared some similarities. The auditory image model [R. D. Patterson et al., in Auditory Physiology and Perception, edited by Y. Casals, L. Demany, and K. Hoerner (Pergamon, Oxford, 1992)] performs an operation similar to autocorrelation, and this temporal model accounts for the pitch and pitch strength of rippled noise and other versions of complex pitch. [Work supported by NIDCD and DRA Farnborough.]