The perceptual organization of noise-vocoded speech under competitive conditions.

Abstract

Research using sine-wave speech suggests that across-formant grouping depends critically on modulation of the frequency but not the amplitude contours of the formants. The generality of this finding was explored using noise-vocoded sentences. Sentences were filtered into six frequency bands, and the amplitude envelope of each band was used to modulate a matched noise-band carrier (N). Bands were paired, corresponding to F1 (=N1+N2), F2 (=N3+N4), and the higher formants (F3′=N5+N6), so that the frequency contour of a formant was implied by variation in the relative amplitudes of the corresponding pair of bands. Perceptual organization was probed by presenting stimuli dichotically (F1+F2C; F2+F3′), where F2C is a competitor for F2 that listeners must resist to optimize intelligibility. F2Cs were derived from F2 by time reversing the envelopes of N3+N4. Implied frequency variation in F2C was manipulated by varying the AM depth in the constituent bands (100%-0% of original depth) or by comodulating both bands with the same envelope (from N3, N4, or N3+N4). Competitor efficacy declined with AM depth. Comodulated competitors were less efficacious than their 100% modulated counterpart. The findings are consistent with the proposal that it is the time-varying frequency contours of formants that govern their grouping. [Work supported by EPSRC.]