Externalization in binaural synthesis: effects of recording environment and measurement procedure

Abstract

Databases of head related impulse responses (HRIRs) for binaural synthesis can be measured either in anechoic or reflective environments. If high synthesis quality is needed, miniature microphone measurements are performed in the ear canals of each individual user (individual measurement). Sometimes impulse responses measured in the ear canals of one individual are used for synthesis for other persons (nonindividual measurement). In most other cases artificial head measurements are used. This paper considers the dependence of the perceived distance of auditory images (externalization) on the measurement procedure (individual, nonindividual, artificial head) and on the recording environment (anechoic, reflective). For each measurement the same system and the same setup, especially the same geometric parameters, are used. Differences in the corresponding impulse response databases are determined and related to the subjective relative externalization differences in the front, in the back, and to both sides. For each direction a seven point rating scale was used. Statistical analysis suggests that the measurement parameters applied influence the externalization of auditory images.