Direction-of-arrival estimation of impulse noise using MUSIC algorithm

Abstract

Spatial information about the source of sounds can be obtained by measuring amplitude and phase differences among multiple microphones in a microphone array. Various methods have been proposed for estimating the direction of arrival, including beam-forming methods, but generally their spatial spectral resolution degrades with a low-frequency source in the far-field. In architectural acoustics, it is important to obtain high spatial resolution in the low-frequency band, especially in investigations to understand phenomena such as sudden impulse noise. In this paper, direction-of-arrival estimation using the MUltiple SIgnal Classification (MUSIC) method is discussed. We propose a microphone array structure designed according to the Cramer-Rao lower limit. It is shown that this technique improves the accuracy of elevation angle estimates at low frequencies while maintaining good horizontal angle estimation accuracy. The proposed method was tested in a real building in which impulse noise occurred and its effectiveness in identifying the source direction with high spatial resolution even at low frequencies was confirmed.