Rapid psychophysical calibration using bisection scaling for individualized control of source elevation in auditory display

Abstract

In an effort to reduce problems stemming from individual differences in spatial hearing, a rapid method for customizing an interactive spatial auditory display for individual users was developed and tested. This paper describes how new users of a DSP-based spatial auditory display system perform a short series of psychophysical calibration tasks via realtime manipulation of the elevation of a virtual sound source removed from the median plane by a constant angle (on a ``cone of confusion'' centered on the interaural axis). The user first produces five settings indicating the point at which the perceived elevation of a virtual source matches their own internal standard for ``ear-level'' incidence. The median of these settings provides an anchoring stimulus for creating an individualized psychophysical scale for controlling source elevation as perceived by the user of the display system. The experimentally-derived anchoring stimulus is regarded as the origin for an angular bisection session that enables the rapid construction of a look up table (LUT) for the full range of elevations produced by the display for each individual user. In contrast to systems that base source elevation control upon individualized head-related transfer functions (HRTFs), the tested system uses a generic set of HRTFs, and manipulates only the values in the LUT for the elevations produced by each HRTF. The method does not attempt to find for each individual listener a single best frequency scaling for the generic set of HRTFs, but attempts to map the useful range of elevations produced by them. Though such a LUT for perceived elevation can be based upon angular estimates made for virtual sources created using each of many HRTFs, the bisection task presented here requires users to complete only a short listening session in which they adjust the elevation of a comparison stimulus to bisect the angle subtended by a pair of reference stimuli. In contrast to other rapid methods of customization, such as those based upon a user's subjective preferences, the current method is based upon active spatial manipulation of a virtual source. The adjustments are referenced to the user's internal standard for ``ear-level'' incidence, which is tangibly defined and quite easily explained to new users.