Auditory motion perception in multiple sound source environments

Abstract

Most everyday listening environments contain multiple sources of sound, and in many contexts, one or more of these sources undergo changes in spatial location that may be perceived as auditory motion – a property of a source that is separate from its fixed, static location. While the basic psychophysics of auditory motion perception have been studied in specific contexts (typically single source “quiet” environments), the usefulness of auditory motion as an independent cue for segregating sound sources has not been established. It also is not known whether source segregation based on perceived motion could form a sufficient basis for selectively attending to one talker among competing talkers in multitalker listening situations. In this dissertation, the basic properties of source motion perception in multisource environments was examined. Because multiple concurrent sources interact to cause masking, the two broad classes of sound source interference – energetic and informational masking – were examined as factors in the perception of source motion. To study perceived motion in multisource settings separate from stationary location differences, a novel experimental paradigm was developed that employed sinusoidal variation implemented by interpolated head-related impulse responses to vary the source location. In the first study, a systematic relationship was established between the amplitude of sinusoidal motion and listener performance on detecting source motion in masked conditions using speech and speech-modulated noise stimuli (i.e., larger motion led to better listener performance). Furthermore, a second experiment demonstrated that target source motion could be used in solving a cocktail party problem (CPP) task. In the second study, listener performance was found to be related to other relevant variables (i.e., speed of source motion and speed of observer response). Although large individual differences were observed in response times for detecting motion, group results indicated that slower motion rates were easier to detect than faster rates consistent with past work describing the “sluggishness” of human binaural processing. Overall, this work established that human listeners can reliably perceive source motion separate from static location differences in multisource settings and can effectively use auditory motion as a cue for segregating and selectively attending to a target source in a multitalker communication situations.