The role of individualized headphone calibration for the generation of high fidelity virtual auditory space

The auditory system localizes a sound source on the basis of binaural cues as well as monaural cues provided by the outer ear. These cues are described by the head-related transfer functions (HRTFs). The individual variation in the frequency and amplitude of perceptually significant features in the human HRTFs raises the issue of whether a given listener can obtain adequate localization cues from stimuli based on another listener's HRTFs. This question can be addressed in "virtual auditory space," using headphone stimuli which are convolved with HRTFs and the inverse of the headphone transfer function (HpTF). Using an in-ear recording system we have measured HpTFs (Sennheiser 250 Linear) for both ears of ten subjects and found that they also display significant individual differences in the 4-10 kHz region. Therefore, the proper reconstruction of both individualized and nonindividualized HRTFs requires that the listener's HpTF be deconvolved from the stimulus. We have verified this acoustically using the in-ear recording system for both individualized and nonindividualized HRTFs. Finally, we show that the use of nonindividualized HpTFs creates considerable distortions in the 4-10 kHz range, which could partly account for the increased incidence of mislocalizations reported in previous virtual auditory space localization experiments.