HRTF modeling in due consideration variable torso reflections

Abstract

Head related transfer functions (HRTFs) describe the physical path from an acoustical source to the ears. It can be gained within the relation of two measurements. The first will give the reference sound pressure in the virtual middle of the head the second has to be done in both ears. In literature exhaustive investigations concerning the idealized location of the second position can be found. Beyond that measurements with various artificial heads and human-beings considering various source directions (azimuth and elevation) with different angular resolutions have been undertaken. The obtained HRTFs exhibit, with subject to the examined source position, frequency dependent magnitude and phase. These complex transfer functions are caused by various reflections, diffractions, pressure superelevation, and shadowing, respectively. Reflections are caused form pinna, head and torso, too. Heretofore during determination of the transfer functions the influence of various head-torso arrangements have not be considered. Within this paper the differences which will be caused due to different head-torso angles will be highlighted. Beyond that a numerical model based on interpolation using spherical harmonics will be presented. By the aid of that model the integration within an existing binaural resynthesis model can be realized and perceptual differences within a dynamic system can be investigated.

Audibility and Interpolation of Head-Above-Torso Orientation in Binaural Technology

“…Our evaluation of the effect of HATO in HRTFs supported findings of earlier studies regarding the comb-filter like nature of the shoulder reflection, which was found to be most prominent if sound source, shoulder, and ear are aligned [4], [6], [7]. Because observed deviations in ITDs and ILDs were below the threshold of audibility for the vast majority of HATOs and source positions, we suppose perceived differences to be mostly due to spectral deviations.…”

“…From an analysis of HRTFs measured for various HATOs, Guldenschuh et al [7] found that the most prominent torso reflections occur when ear, shoulder, and source are approximately aligned, and the source elevation is within 20 • below the horizontal plane to 40 • above. The authors further hypothesized that effects caused by the torso should be audible at least for critical source positions.…”

Spatial Audio Feature Discovery with Convolutional Neural Networks

“…These maps indicate that model WB uses filter 2 to extract ripples in the range from 0.3 to 4 kHz caused by shoulder and torso reflections [15]. One limitation of the datasets used in this study is that the dependence of shoulder reflections on head orientation [29] is not accounted for. Training the model on variable shoulder reflections might potentially lower the contribution of these cues.…”

Measurement of Head-Related Transfer Functions: A Review

“…The torso can act as a shield or a reflector depending on the HATO and source directions, and shows noticeable influences on measured HRTFs [185,208]. Brinkmann et al [209] demonstrated the audible deviations of HRTFs measured with fixed and variable HATOs.…”