Improved understanding of head-related transfer functions through sagittal-plane spectral decomposition

Abstract

Head-related transfer functions (HRTFs) capture the listener-specific modifications that are imparted on a sound as it travels from a specific location in space, interacts with the listener’s head, shoulders, and outer ears, and arrives at the ear drums. HRTFs have often been decomposed into a combination of two parts to aid modeling and analysis: a diffuse-field transfer function (DFTF) that captures non-directional but still listener-specific spectral details of the HRTF, and a directional transfer function (DTF) that captures all of the listener-specific directional information. Because the DTF contains spectral variation along all spatial dimensions, it does not provide a way to separate the spectral cues responsible for lateral and sagittal plane localization judgments, which are largely independent. As an alternative approach, decomposing an HRTF along each sagittal plane (aka along the “cones-of-confusion”) provides similar benefits to DTF analysis while providing additional insights into the spectral cues responsible for lateral and vertical localization, independently. This talk will provide a detailed overview of sagittal-plane spectral decomposition along with experimental results showing how the technique can be applied to improve localization with non-individualized HRTFs, separate general and idiosyncratic HRTF features, and cluster listeners’ HRTFs based on spectral similarity.