Analysis and synthesis of three-dimensional sound fields using a hybrid microphone array and a 3-layered loudspeaker array

Abstract

An integrated recording-rendering array system is constructed for three-dimensional reverberant environments. The array system is comprised of a hybrid microphone array consisting of a circular array with vertical extension for recording and a 3-layered loudspeaker array playback system. In the sound field analysis (SFA) stage, the hybrid microphone array is used to encode the sound field on the basis of plane-wave decomposition. The bearing and amplitude information of the plane waves can be retrieved using either a one-stage Newton's (NT) algorithm or a two-stage method. In the two-stage method, localization can be handled using either the minimum power distortion response beamformer or the multiple signal classification algorithms, followed by signal separation using either Tikhonov regularization (TIKR) algorithm or the NT method. In the sound field synthesis stage, to construct the room response model as required in field matching process, uniformly distributed plane waves are assumed. A two-step procedure is employed to select the regularization parameters in the TIKR algorithm to strike a balance between reconstruction performance and audio quality. A 3-layered loudspeaker array is implemented for reproducing the sound field that is parameterized previously in the SFA stage. The proposed acoustic array system is validated in a series of localization and listening tests.