Abstract:

Neff and Green [Percept. Psychophys. 41, 409--415 (1987)] reported that the masking of single tones by random-frequency multitone maskers is greatest when the masker is comprised of between 10 and 50 tonal components. In this paper we present data to suggest that such results are related to the frequency spacing of masker components. Two conditions were run. In the first, the number of tones comprising the masker was varied from 20 to 906, with frequencies chosen at random over a fixed range from 0.1 to 10 kHz. In the second, the number of tones was fixed at 20, and the frequency range over which components were chosen was varied to match the frequency spacing of components in the first condition. An adaptive two-interval, forced-choice procedure was used to measure masked thresholds for a 1.0-kHz signal. Both conditions yielded large elevations in signal threshold. This was true even after compensating for the effects of energetic masking produced by components nearest the signal. Compensated thresholds showed little dependency on component number, but decreased systematically with decreasing frequency spacing of components. The results are consistent with a model of multitone masking based on component-relative entropy [R. A. Lutfi, J. Acoust. Soc. Am. 94, 748--758 (1993)].