Stanley Sheft
William A. Yost
Parmly Hearing Inst., Loyola Univ., 6525 N. Sheridan Rd., Chicago, IL 60626
In a 2IFC paradigm, listeners indicated which of two concurrent harmonic complexes was amplitude modulated (AM). The fundamental of one complex was always 200 Hz and the fundamental of the other was either 220 or 245 Hz. Stimuli were three randomly chosen harmonics of both fundamentals. The harmonics of one fundamental were modulated in the first interval and those of the other in the second interval. The AM rate was either 5, 13, or 23 Hz. In control conditions the two harmonic complexes were presented sequentially in random order. Harmonic randomization significantly affected performance in all conditions. Even in conditions with the lowest variability (selecting three harmonics from a pool of four or five), performance levels for some subjects were below a d' of 1.5. With coherent modulation in the concurrent pitch segregation task, performance came closest to the control conditions with the 45-Hz f[sub 0] separation, 5-Hz AM, and low variability in harmonic randomization. Incoherent modulation introduced by randomizing either modulator phase or rate made the task more difficult as did the addition of flanking tones unrelated to either f[sub 0]. Results indicate that though dependent on stimulus variability, coherent AM can lead to spectral grouping for pitch segregation. [Work supported by NSF and AFOSR.]