Using vocoded speech to study temporal weighting in spatial release from masking

Abstract

Several aspects of binaural and spatial hearing appear to be dominated by moments of rising envelope fluctuations (“onsets”), although those phenomena are limited to events occurring less than a few hundred times per second. Thus, for example, the localization of a brief tone is dominated by the binaural information available in the overall sound onset. Whereas for a modulated sound each envelope period may contribute equally, assuming the modulation rate is not too high. For speech signals, syllabic modulations provide multiple onset-like events which may occur synchronously or asynchronously across frequency bands. Although previous studies have shown that rising envelope fluctuations play an outsized role in localization and lateralization, it is unknown whether other aspects of spatial hearing–such as spatial release from masking (SRM)–are similarly dominated by rising envelopes. Here, we used a Gabor-click-train vocoder to transform speech sounds and manipulate the spatial content of rising- versus falling-envelope segments of the speech in each frequency band. Introducing binaural differences between target and masker speech allows the assessment of localization and/or SRM across conditions in which the cues are limited to rising- or falling-envelope segments, or available throughout the signal. [Work supported by NIH R01DC016643, T32000013.]