Computing HRTFs using Spherical Near field Acoustic Holography for rendering spatial audio

Abstract

In this paper, a novel method for computing Head Related Transfer Functions (HRTFs) using principles of Spherical Nearfield Acoustic Holography (SNAH) is described. To compute generalized HRTFs the head and the torso have been modeled by spheres of different radii using the snowman model. A new method of sound pressure reconstruction over a spherical surface that utilizes SNAH is proposed for modeling the head. Pressure reconstruction corresponding to all spatial locations of a sound source is performed using motion holography. The torso is modeled by using principles of Gaussian optics within a SNAH framework. A combined head and torso (HAT) model is then developed to obtain HRTFs. Head and Torso response obtained by the proposed approach is compared individually with existing numerical techniques. For low frequencies both approaches provide similar results. Binaural audio is then synthesized using HRTFs computed from the proposed method and from the CIPIC database. Subjective evaluation on binaural audio convey that proposed technique results are comparable to CIPIC in both horizontal and frontal planes.