Estimation of Hearing Aid Head Related Transfer Functions using Anthropometric Features

Abstract

Virtual reality is emerging as an ecologically valid environment for auditory rehabilitation and diagnostics. Creating an auditory environment that accurately simulates a user’s experience requires accurate knowledge of their Head-Related Transfer Functions (HRTFs). However, these vary due to differences in physical features and can lead to the incorrect simulation of sound if the HRTFs used are too dissimilar to a listener’s. This issue is exacerbated by the use of a hearing device as sound is transmitted to the eardrum from a microphone located elsewhere on the ear, changing the path of sound and consequently the HRTFs. This project presents a method of identifying a suitable set of HRTFs from a dataset, performing better than chance and leads to a localisation error 1.2◦greater than a subject using their HRTFs.