Auditory vertical localization in the median plane with conflicting spectral and dynamic cues

Abstract

Binaural pressures include the cues for auditory localization. The scattering and diffraction of the anatomical structures such as head, pinna and torso modify the pressure spectra at the ears (spectral cue). Head turning changes the binaural pressures (dynamic cue). Both spectral and dynamic cues provide information for front-back and vertical localization. Auditory vertical localization is the consequence of comprehensive processing of multiple localization cues by the high level nervous system. In the case of a real sound source, multiple localization cues provide consistent information, enhancing the accuracy in vertical localization. In some cases of spatial sound reproduction, however, some vertical localization cues may be absent (such as in the case of stable virtual auditory display), error, or even conflicting. Therefore, it is necessary to examine the contribution of spectral and dynamic cues to vertical localization. However, the relative importance of two cues is still unclear. The spectral cue has conventionally been regarded as a dominant one and the contribution of dynamic cue was often neglected. In present work, a psychoacoustic experiment was conducted to investigate the vertical localization in the median plane with conflicting spectral and dynamic cues, so as to examine the relative importance of two cues in vertical localization. A dynamic virtual auditory display (VAD), which was able to detect the three degrees of freedom of head turning and then stimulate the binaural signals as well as their variation with head turning, was used as a tool in the experiment. By modifying the head-related transfer functions (HRTFs) used in the dynamic VAD, binaural signals with conflicting spectral cue and dynamic variation of interaural time difference (ITD) were created. That is, for a target source in the median plane, the magnitudes of binaural signals and their variation with head turning were simulated as if target source was located at an elevation (called spectra-based elevation); while the ITD and its variation with head turning were simulated as if target source is located at another elevation (called ITD variation-based elevation). The resultant binaural signals were presented to subjects by a pair of headphone. The subjects reported the perceived virtual source directions. Eight subjects with normal hearing participated in the experiment. A statistical analysis was applied to the experimental data from all subjects. The statistics of experimental results indicated that, for low-pass filtered noise with cut-off frequency of 3 kHz, the perceived virtual source elevation varies basically consistent with dynamic ITD variation (ITD variation-based elevation), regardless of the spectral cue. For pink noise with full audible bandwidth, conflicting spectral and dynamic cues result in two splitting auditory events or perceived virtual sources at different elevations, rather than a fused auditory event at an elevation. The perceived elevation of the virtual source corresponding to the low-frequency component of stimulus is basically consistent with dynamic ITD variation (ITD variation-based elevation), while the perceived elevation of another virtual source correspond to the high-frequency component of stimulus is basically consistent with the spectral cue (spectra-based elevation). Therefore, the present experiment proved that both dynamic and spectral cues are important to vertical localization. Dynamic cue dominates the vertical localization below the frequency of 3 kHz, and the spectral cue dominates the vertical localization above the frequency of 3 kHz.