HRTF personalization based on ear morphology

PubDate: Aug 2022

Teams: Reality Labs Research

Writers: Michaela Warnecke, Sharon Jamison, Sebastian Prepeliță, Paul Calamia, Vamsi Krishna Ithapu

PDF: HRTF personalization based on ear morphology

Abstract

On the forefront of realistic spatial audio is the personalization of binaural auditory input. Specifically, personalized head-related transfer functions (HRTFs) have been shown to improve the quality of binaural spatial audio over generic HRTFs. Here, we approached HRTF personalization from a morphological standpoint by calculating the distance between any two three-dimensional models of the ear. Subsequently, a ranking of ears based on the distances provided a similarity estimate. Using measured HRTFs of these ears, we tested how well listeners performed when localizing sounds in virtual space. We show that performance is closest to that of the individual HRTF when listeners used the best-ranked ear’s HRTF and furthest when listeners used the worst-ranked ear’s HRTF.