Investigation Of Sound Reproduction Systems Based On Loudspeaker Array

The development of virtual reality(VR)technology requires increasingly higher performance of sound reproduction systems.Spatial sound reproduction systems using loudspeaker array have drawn attention from both scholars and engineers.They can be generally divided into two categories:(1)perceptual reconstruction of the stereo sound scene,and(2)physical reconstruction of acoustic field.For the perceptual sound scene reconstruction,appropriate crosstalk cancellation algorithms are required to guarantee accurate binaural sound reproduction.Crosstalk cancellation commonly used in the two-channel reproduction systems usually faces the problem of severe spectral coloration effects.In the first part of this paper,two different regularization methods alleviating this problem based on measured transfer functions are investigated.Experiments are carried out to compare the spectral coloration alleviation and the crosstalk cancellation performance.When more than two loudspeakers are utilized,optimal source distribution(OSD)strategy is a proper method optimizing the source positions and the corresponding playback frequency.It has the benefit of achieving great performance within the whole frequency range of audible sound.However,the performance deteriorates considerably when the head is laterally misaligned.In the second part of this paper,an improvement method for a linear binaural playback system is proposed,in which the OSD strategy in a linear array configuration is combined with the technique of multiple control points.Both the simulation and experiment results verify its strong robustness against head misalignment.Ambsonics is a strategy to reconstruct sound field physically.The most simplified Ambisonics systems render 3D sound field from the omnidirectional and the first order directive information,known as the B-format.Higher-order Ambisonics(HOA)extends B-format to higher spatial resolution by spherical harmonic decomposition of the sound field.For the past three decades,massive amount of studies have been conducted to investigate the accuracy and limitation of the HOA though few of them have taken into account the diffraction of the human head and the features of the human auditory system.In the third part of this paper,the sweet-zone size of different order HOA systems is evaluated in terms of the interaural time difference(ITD).The range of sweet-zone increases markedly with higher order Ambionics system,which manifests that HOA has better robustness when human head is located away from the center of the listening zone.