Simple Visual Cues Enhance the Identification of Target Sounds in Complex Auditory Scenes

Erol Ozmeral, Auditory Neuroscience Laboratory, Boston University, Boston USA

Abstract
We investigated the extent to which non-auditory cues for directing auditory attention improved the ability to identify an acoustic target embedded in a complex auditory scene. Two types of stimuli were used: 1) sequences of spoken digits with reversed digits as distracters and 2) familiar Zebra Finch songs with unfamiliar songs as distracters. Subjects were seated in front of five loudspeakers (-40° to +40° azimuth, 1 m distance), each mounted with an LED. Five different streams played simultaneously from the five loudspeakers. The listener identified a target, which occurred from one of the loudspeakers at a random time. There were four conditions: 1) all LEDs turned on during the target time-slice (“when” cue), 2) the target LED remained on throughout the trial (“where” cue), 3) the target LED turned on during the target time-slice only (“when and where” cue), and 4) no LEDs turned on (“no cue”). “When and where” cues yielded the best performance; “no cue” produced the worst; the other conditions were intermediate. The results demonstrate that in a complex auditory scene, a simple visual cue can improve the ability of listeners to extract information about an auditory target. [Work supported by grants from AFOSR & ONR]

Not available

Back to Abstract