The influence of informational masking in complex real-world environments

Spatial release from masking (SRM) is believed to be an essential auditory mechanism aiding listeners in reverberant multi-talker environments. However, SRM is often measured in simplified spatial configurations using speech corpora with exaggerated talker and/or context confusions. Besides energetic better-ear listening and binaural unmasking, the perceived spatial separation of target and masking speech signals is thought to aid listener's segregation of speech signals, resulting in a so-called release from informational masking. This study aims to estimate the amount of informational masking that is apparent in complex real-world environments. Speech reception thresholds (SRTs) were measured by presenting Bamford-Kowal-Bench (BKB) sentences in a simulated cafeteria environment recreated by a spherical array of 41 loudspeakers placed in an anechoic chamber. Three maskers with varying degree of informational masking were realized: one with talkers different from the target, one with an unintelligible noise vocoder (minimal informational masking) and one with the same talker as the target (maximum informational masking). The maskers were constructed with either two or seven two-talker conversations and were either spatially distributed in the simulated cafeteria or colocated with the target. Seven normal hearing listeners were tested. All conditions showed improved thresholds for the spatialized condition compared to the colocated condition. However there was no significant difference between the different talker speech and vocoded masker. Only the same talker masker showed increased thresholds and this was only substantial in the two conversation colocated condition. These results suggest that informational masking is of low relevance in real-life listening and is exaggerated in listening tests by target/masker similarities and the colocated spatial configuration. However, this may be different in (aided) hearing impaired listeners where spectral and spatial cues can be significantly disturbed.