Individual equalization in binaural reproduction of a reverberant sound field

Abstract

Currently, there is a constraint in the individual equalization of the differences in head-related transfer functions (HRTFs) between a dummy head and a listener for a binaural system. Therefore, the equalization function of a listener must be measured in the original sound field. To remove this constraint, we have developed a method that replaces the function measured in the original sound field with that measured in another sound field (an equalization sound field). The two functions’ profiles are similar. Therefore, the original function is approximated by adding reflected sound components to the transfer function measured in an equalization sound field. In the proposed method, the reflected sound components are extracted by applying the auto regressive moving-average (ARMA) model to the transfer function measured in the original sound field. An ear canal transfer function is introduced into the proposed method to calculate the transfer function measured at the entrance of the ear canal from that at the eardrum. A psychoacoustical experiment is carried out to evaluate the equalized sound with respect to four criteria and localization. The experimental results confirm that the proposed method is effective for all listeners.