Binaural range estimation using Head Related Transfer Functions

Abstract

Inspired by the role of the outer ear to shape, direct, and focus sound waves, we present a binaural method for sound source range estimation in a three dimensional space to be deployed in a humanoid hearing system. The method is designed to allow robots to localize sound sources using only four microphones, two of which are placed inside the ear canals of a humanoid head equipped with artificial ears and mounted on a torso. The algorithm relies on recovering important cues of the human binaural auditory system, primarily encapsulated within the heard related transfer function (HRTF). Based on the content of the collected sound signals, the distance between the microphones reconfigures automatically in order to optimize the localization accuracy. While existing 3D sound source distance estimation techniques use microphone arrays, the presented method employs only four microphones and is based on a simple correlation mechanism using a generic set of HRTFs.