I'm excited to announce that our paper titled 'A Numerical Study into Perceptually-Weighted Spectral Differences between Differently-Spaced HRTFs' has been accepted for the 2019 AES International Conference on Immersive and Interactive Audio in York in March.
We were interested in the perceptual impact of using HRTFs which are slightly different to the intended ones - particularly when this is due to the application. Ambisonic rendering requires specific angles, often to 2 decimal places - what is the impact of using HRTFs measured at a slightly different angle? Often, the ears are the point of interest, but HRTFs are measured relative to the centre of the head - what impact does this change of origin have? Do these differences become negligible at a certain distance?
We simulated HRTFs for KEMAR using BEM (with Mesh2HRTF) at a high number of radial distances and using three origins - the centre of the head, and the position of each ear - and compared the differences between them using a perceptual difference model. We found that average variation reaches perceptual limits by 2-3m, however, there remained perceivable differences up to and beyond 10m, indicating that care must be taken when approximating either the distance or angular location of HRTFs.
Update: the paper is available here.
Cover photo by Lubos Houska from Pixabay.
Comments