Individualized and Generalized Earphone Correction Filters for Spatial Sound Reproduction

Earphone-based simulation of Head-Related Transfer Functions (HRTFs) is a common component of spatial auditory display for virtual sound sources located in virtual environments, but one technical component of HRTF deployment that requires special care has not always received proper attention. The audio signals to be presented to the listener are not often properly corrected for the response of earphones, and yet such correction has long been regarded as crucial to the success of spatial sound reproduction using earphones. Two approaches to developing Earphone Correction Filters (ECFs) are contrasted in this paper, one based on the assumption that the ECF must be based upon an inversion the user’s individually measured Earphone Transfer Functions, and the other based upon the assumption that a satisfactory result can be attained through a general equalization for correcting the tone coloration imparted by the averaged ETF. The results of these two approaches were perceptually evaluated for a range of stimuli generated by varying the shape of an experimental ECF using a fouralternative, forced-choice (4AFC) test. Changes in auditory imagery associated with speech stimuli were easily detected and discriminated when comparing reproductions with and without earphone correction, but perceptual differences were much more difficult to detect as the shape of the experimental ECF was varied by frequency scaling. Further forced-choice tests showed that ECF frequency scaling factors of 0.8 and 1.2 do not result in detectable differences, but factors of 0.6 or 1.4 make detectable differences.