Threshold differences for interaural time delays carried by double vowels.

Abstract

Experimental measurements were made of threshold interaural time differences (ITDs) for a "target" vowel presented simultaneously with a fixed-ITD "distracter" vowel. Three double-vowel pairs were used, comprising an "er" (/e/) together with either an "ai," "ar," or "oo" (respectively, /e/, /c/, and /u/). Threshold ITDs were found to be larger for the target vowel when it was part of a double-vowel pair than in control conditions in which it was presented alone. The effect size depended upon the choice of target vowel and distracter vowel, the level of the target relative to the distracter, and whether the two vowels had the same or different fundamental frequencies. The experiment was analyzed using a multichannel modification of Heller and Trahiotis' [J. Acoust. Soc. Am. 99, 3632-3637 (1996)] model, which used a weighted combination of the detectabilities of the ITD of the target and the distracter. It gave predictions consistent with the observed effects of level and with some of the effects of the choice of target vowel, but it could not describe the effect of the target-distracter differences in fundamental frequency. It was found that a single-channel version of the model, in which the chosen channel was allowed to depend upon fundamental frequency (which could be derived using a monaural autocorrelation model) did give a set of predictions in qualitative accord with the data.