Comparison of Acoustic Source Localization Methods in Time Domain Using Sparsity Constraints

This paper deals with source localization techniques in time domain for broadband acoustic sources. The goal is to detect accurately and quickly the position and amplitude of noise sources in workplaces in order to prevent employees from hearing loss or safety risk. First, the generalized cross correlation associated with a spherical microphone array is used to get a raw noise source map. Then a linear inverse problem is defined. Commonly, linear inverse problem is solved with an l2-regularization. In this study, two sparsity constraints are used to solve the inverse problem, the orthogonal matching pursuit and the truncated Newton interior-point method. Synthetic data are used to highlight the abilities of such techniques. High resolution can be achieved for various acoustic sources configurations. Moreover, the amplitudes of the acoustic sources are correctly estimated. Finally, a comparison of computation time shows these techniques are suitable in real scenario.