Two Are Better Than One: Solving the Problem of Vertical Sound Source Localization via Binaural Integration of HRTFs

Abstract

Early studies have shown that the localization of a sound source in the vertical plane can be accomplished with only a single ear, thus assumed the localization mechanism to be based on monaural cues. Such cues are induced by the pinna and consist of notches and peaks in the perceived spectrum which vary systematically with the elevation of sound sources. These processes pose several problems to the auditory system like identifying and extracting spectral cues on a neural level, as well as, distinguishing pinna induced peaks and notches from features already present in the source spectrum. Interestingly, at the stage of elevation estimate binaural information from both ears is already available and it seems plausible that the auditory system takes advantage of this information. Especially, since such a binaural integration can improve the localization performance dramatically as we demonstrate in the current study. For that, we first introduce a computational model architecture that takes advantage of binaural signal integration to localize sound sources in the median plane. Model performance is tested under different conditions which reveal that localization of monaural, as well as binaural inputs is best when the model is trained with binaural inputs. Furthermore, modeling results lead to the hypothesis that sound type specific prior information is taken into account to further improve localization quality. This deduced hypothesis about vertical sound source localization is confirmed in a behavioral experiment. Based on these results, we propose that elevation estimation of sound sources is facilitated by an early binaural signal integration and can incorporate sound type specific prior information for higher accuracy.