Speech intelligibility and spatial release from masking in maskers with different spectro-temporal modulations

Abstract

Speech-reception thresholds (SRTs) decrease as target and maskers are spatially separated (spatial release from masking, SRM) even if two maskers are symmetrically placed around the listener's head. In this case, speech intelligibility (SI) cannot be explained by an improved long-term signal-to-noise-ratio (SNR) caused by the head shadow at one better ear alone, but could be facilitated by short-term spectro-temporal segments (“glimpses”) in each ear that provide favorable SNRs. Here it was systematically assessed how SRT and SRM depend on the spectro-temporal masker properties and on the availability of specific binaural cues for a frontal target in a symmetric masker setup. Maskers ranged from stationary noise to single, interfering talkers. Maskers were modified by head-related transfer functions providing different binaural cues (interaural level and time differences; ILD, ITD, both), by presenting only glimpses derived with a fast-switching better-ear mechanism, and an “infinite ILD,” removing crosstalk of the maskers between the ears. Results were compared to model predictions showing that spectral cues contribute to SRM for all maskers, while IPD and ILD cues were more important for modulated maskers. The “infinite ILD” condition suggests binaural processing limitations, resulting in a maximal SRM of 12 dB for low or absent informational masking.