Influence of acoustic complexity on spatial release from masking and lateralization

In realistic listening scenarios, humans are extremely skillful in following one particular talker even in the presence of many others (i.e., the cocktail party effect). One aspect of this is the ability of a listener to make use of the spatial separation between different sound sources. In a complex acoustic scene, as interferers are moved away from the spatial position of the target, speech intelligibility (SI) increases, often referred to as spatial release from masking (SRM).

This benefit is largely based on the listeners’ ability to make use of interaural level differences (ILD) and interaural time differences (ITD), which vary with the source location. While many studies have explored SRM, few have investigated the effects of overall number and spatial distribution of interferers while controlling for monaural masking effects.

In the present study, speech reception thresholds (SRTs) and lateralization thresholds were measured over headphones in babble noise conditions consisting of 2, 4, 8 and 12 talkers. The perceived locations of the signal (female voice) and individual maskers (male, time-reversed voices) were steered separately either to the left or to the right using 680 sec ITDs.

For a fixed number of maskers, the distribution of interfering talkers (i.e., co-located or separated from the target) was varied. Thus, for all conditions with the same number of maskers, the monaural SNR was held constant regardless of the perceived spatial distribution. The performance between the speech and lateralization tasks was highly correlated.

No substantial SRM occurred while one or more maskers were co-located with the target. Interestingly, spatially shifting the last 2 co-located interferers resulted in the same SRM, independent of the overall number of maskers. The same finding was found for the last 4 co-located maskers. The results suggest that SRM is independent of the overall number of interfering talkers.