Investigation of ITD Symmetry in Measured HRIRs
Genovese, Andrea F.
MetadataShow full item record
The Interaural Time Difference is one of the primary localization cues for 3D sound. However, due to differences in head and ear anthropometry across the population, ITDs related to a sound source at a given location around the head will differ from subject to subject. Furthermore, most individuals do not possess symmetrical traits between the left and right pinnae. This fact may cause an angle-dependent ITD asymmetry between locations mirrored across the left and right hemispheres. This paper describes an exploratory analysis performed on publicly available databases of individually measured HRIRs. The analysis was first performed separately for each dataset in order to explore the impact of different formats and measurement techniques, and then on pooled sets of repositories, in order to obtain statistical information closer to the population values. Asymmetry in ITDs was found to be consistently more prominent in the rear-lateral angles (approximately between 90° and 130° azimuth) across all databases investigated, suggesting the presence of a sensitive region. A significant difference between the peak asymmetry values and the average asymmetry across all angles was found on three out of four examined datasets. These results were further explored by pooling the datasets together, which revealed an asymmetry peak at 110° that also showed significance. Moreover, it was found that within the region of sensitivity the difference between specular ITDs exceeds the just noticeable difference values for perceptual discrimination at all frequency bands. These findings validate the statistical presence of ITD asymmetry in public datasets of individual HRIRs and identify a significant, perceptually-relevant, region of increased asymmetry. Details of these results are of interest for HRIR modeling and personalization techniques, which should consider implementing compensation for asymmetric ITDs when aiming for perceptually accurate binaural displays. This work is part of a larger study aimed at binaural-audio personalization and user-characterization through non-invasive techniques.