Release of masking for speech through interaural time delay.

Abstract

Interference with binaural intelligibility of spondaic words produced by continuous white noise and of monosyllabic words produced by both continuous and modulated white noise as well as connected speech (single talker) was studied under a variety of interaural listening conditions. Performance during homophasic (N0S0) and antiphasic (NπS0) listening was compared with that achieved under conditions involving various interaural time differences of the noise and/or the speech. These time differences ranged from 0.1 to 0.8 msec. Several signal-to-masker ratios were employed, but for the conditions involving modulated noise, only two modulation rates (4/sec and 100/sec) and a single magnitude of modulation (14 dB) were used. Transition from homophasic to antiphasic listening produced masking-level differences (MLD's) of about 7 and 4 dB for spondees and monosyllables, respectively. The MLD's produced by varying the interaural timing of either speech or noise increased systematically as the time differences were increased within the range studied here, but they never exceeded those for antiphasic listening and were usually appreciably smaller. As gauged by performance under 0.4- and 0.8-msec interaural time delay, the MLD for monosyllables was the same regardless of whether the time difference was applied to the masker or the speech. Furthermore, the MLD produced by simultaneous but opposing interaural time differences (masker leading in one ear and speech in the other) did not exceed the antiphasic MLD, even when the aggregate timing discrepancy between the two signals reached 1.6 msec. Some implications of these findings are discussed.