Computational quality assessment of HRTFS

The verification of quality in head-related transfer function (HRTF) measurement data holds a complex problem field, involving acoustic (background noise), physical (head movement, different head shape) and electrical (DC-offset) deviations. The disrupted measurements may be difficult to separate from the correct ones, because HRTFs are inherently highly dependent on the person, direction, and frequency. Since there exists no “absolute truth” what a HRTF should be like, deviated measurements can be assessed from the typical HRTF shape, i.e., the common trend of the measured HRTFs. This paper presents an approach to calculate an objective deviant measure of HRTFs by DSP means. The perceptually justified deviance index is applied for the evaluation of a HRTF measurement system and a HRTF database.