Deconvolution using CLEAN-SC for acoustic source identification with spherical microphone arrays. (3rd February 2019)

• Record Type: Journal Article
• Authors: Chu, Zhigang; Zhao, Shuyi; Yang, Yang; Yang, Yongxin
• Is Part Of: Journal of sound and vibration. Volume 440(2019)
• Page Start: 161
• Page End: 173
• Publication Date: 2019-02-03
• Subjects: Acoustic source identification -- Solid spherical microphone array -- Deconvolution -- CLEAN-SC; Sound -- Periodicals; Vibration -- Periodicals; Son -- Périodiques; Vibration -- Périodiques; Sound; Vibration; Periodicals; Electronic journals; 620.205
• DOI: 10.1016/j.jsv.2018.10.030
• Languages: English
• ISSNs: 0022-460X

Abstract

Abstract: The point spread function (PSF) based deconvolution algorithms may fall short due to the inconformity between the actual beam pattern of acoustic sources and the theoretical PSF. CLEAN-SC, originally developed for delay and sum beamforming with planar microphone arrays, can circumvent the inconformity. This paper devotes to adapting CLEAN-SC to spherical harmonics beamforming (SHB). The core of adapting CLEAN-SC to SHB is to express the output of SHB in the spatial domain as a specific form of matrix operation completely. We solve the issue and establish CLEAN-SC with spherical microphone arrays. Its performance is analyzed and compared with PSF based CLEAN. CLEAN-SC with spherical microphone arrays can improve spatial resolution and suppress sidelobes more effectively, as well as quantify the sound pressure contribution accurately. Moreover, it has better convergence, higher computational efficiency, and stronger robustness to interference such as background noise and frequency response mismatch of microphone and measurement channel. This study provides an alternative approach for acoustic source identification in three-dimensional space. Highlights: CLEAN-SC is adapted to SHB with spherical microphone arrays. It can identify the acoustic sources accurately. It has obvious advantages over SHB in resolution and dynamic range. It has faster computation, better convergence, and stronger robustness than CLEAN.