The effect of stimulus bandwidth and subject position on horizontal-plane localization with virtual source images

Abstract

In an anechoic chamber normal-hearing subjects performed a localization task in the frontal horizontal plane. The stimulus was a 200-ms burst of filtered noise. Within a block of trials, half of the presentations (randomly determined) were ‘‘real’’—presented from single loudspeakers—and the other half were ‘‘phantoms’’—produced by the simultaneous activation of two loudspeakers at ±30° using a virtual source imaging technique [Takeuchi et al., J. Acoust. Soc. Am. 109, 958–971 (2001)]. Both phantom and real sources spanned the azimuthal range ±80°. When the stimulus was a 4 kHz low-pass filtered noise, rms error was only slighly higher for phantom (D̄=7.1°) than for real (D̄=5.5°) sources. For 8 kHz low-pass filtered noise, performance remained about the same for real sources, but increased for phantom sources (D̄=11.5°). Data will also be reported for conditions in which the subject’s position is systematically varied outside the ‘‘sweet spot.’’ Results will be discussed in terms of robustness of the virtual imaging technique to stimulus and position factors and its potential usefulness as a tool for the investigation of human auditory spatial perception in static and dynamic environments. [Work supported by NIDCD.]