Effects of adding a spectral peak generated by the second pinna resonance to a parametric model of head-related transfer functions on upper median plane sound localization

The parametric head-related transfer function (HRTF) recomposed of only a spectral peak (P1) and two spectral notches (N1 and N2), which are respectively generated by the first resonance and the first and second anti-resonances of the pinna, has been reported to provide approximately the same localization performance as the measured HRTF for the front and rear directions. However, for the upper direction, the localization performance for some of the subjects decreased. In the present study, we conducted two localization tests with four listeners and seven target angles in the upper median plane (0-180°) to investigate whether adding a spectral peak (P2), generated by the second resonance of the pinna, can resolve this performance decrease. The results suggested that (1) the mean vertical localization error of the parametric HRTF recomposed of N1, N2, and P1 was significantly larger than that of the measured HRTFs at the target vertical angles of 30° and 120°; (2) by adding P2 to N1N2P1, the mean vertical localization error decreased at the target vertical angles of 0°, 30°, 90°, and 120°, and no statistically significant difference was observed between N1N2P1 + P2 and the measured HRTFs at any target vertical angle; and (3) a sound image was hardly perceived in the upper direction by reproducing only P2, but the presence of P2 to improve the salience of N1 was discussed.