Detection and localization of speech in the presence of competing speech signals

Abstract

Auditory displays are often used to convey important information in complex operational environments. One problem with these displays is that potentially critical information can be corrupted or lost when multiple warning sounds are presented at the same time. In this experiment, we examined a listener's ability to detect and localize a target speech token in the presence of from 1 to 5 simultaneous competing speech tokens. Two conditions were examined: a condition in which all of the speech tokens were presented from the same location (the `co-located' condition) and a condition in which the speech tokens were presented from different random locations (the `spatially separated' condition). The results suggest that both detection and localization degrade as the number of competing sounds increases. However, the changes in detection performance were found to be surprisingly small and there appeared to be little or no benefit of spatial separation for detection. Localization, on the other hand, was found to degrade substantially and systematically as the number of competing speech tokens increased. Overall, these results suggest that listeners are able to extract substantial information from these speech tokens even when the target is presented with 5 competing simultaneous sounds.