Auditory evaluation of head-related transfer functions from multiple databases

Abstract

In order to rationalize the combined utilization of multiple databases of head-related transfer functions (HRTFs), this work investigates the influence of different HRTF measuring conditions on auditory perception. First, the HRTFs of KEMAR artificial head from five databases were selected, and the diffusion-field equalization was applied to the original HRTFs as a pre-processing method to reduce the differences from different measuring conditions. Then, the spectral deviations among the equalized HRTFs were calculated to objectively evaluate the HRTF’s differences from different measuring conditions. Finally, a series of listening experiments (including localization and discrimination experiments) were conducted to subjectively evaluate different HRTF measuring conditions on auditory perception. Results indicate that the different measuring conditions has less influence on the localization performance of HRTFs, suggesting the auditory consistency among multiple HRTF databases in terms of localization; however, only part of HRTF databases are consistent in terms of indiscernibility which is a stricter metric. [Work supported by National Natural Science Foundation of China, No.11474103.]