Auditory Spatial Cuing for Speech Perception in a Dynamic Multi-talker Environment

Abstract

An experiment was performed to measure how dynamic changes in a target talker location affect speech perception in a complex multi-talker environment. The listener's task was to report a number sequence spoken by the target talker, masked by four distractor talkers differing in their voices and spatial locations. The location of the target was either fixed during a trial (static condition) or randomly changing from one number to the next one (dynamic condition). The distractor talker locations were always randomized between numbers. A cue word sequence spoken by the target speaker with no distractors preceded each trial. The cue indicated the voice (dynamic condition) or the voice and location (static condition) of the target. A large decrease in performance was observed in the dynamic condition compared to the static condition. The position of the target number in a sequence, as well as the inter-word interval, only had a small effect on performance. The decrease in performance in the dynamic condition is either due to a cost of switching attention from one location to another, which can be large when attention is controlled by audition, or due to a change in listeners' strategy from selective focusing to one location (in static condition) to attempting to process all stimuli concurrently (in the dynamic condition).