Beamforming in near-field - metaheuristic approach and experimental tests in an anechoic chamber

A set of microphones spatially arranged in space in a specific pattern is called a microphone array. It can be used to extract and enhance the signal of interest from its observation corrupted by other interfering signals, such as noise or to estimate the direction of arrival of a source. In this paper we focus on a problem in which the desired signal (speech signal) is interfered by other signal with fully overlapping bandwidth but with different localization. Our goal is to attenuate the interfering signal. We experimentally study the method in which microphones do not have to be equally spaced and all information regarding signal phase is hidden in a transfer function of the microphone. We focus on determining the microphones positions and FIR filter coefficients so that the actual output the beamformer is as close as possible to the desired one (the level of speech signal remains unchanged, while the interfering signal is attenuated) in the sense of l2 norm. To solve this problem, we use a metaheuristic algorithm. Next, we construct the array and make an experiment in anechoic chamber. The initial results of the experiment show that the proposed method can be applied for array designing.