Power estimation of acoustical sources by an array of microphones

International audienceThe problem of estimating the strengths of signals arriving at an array of microphones when the arrival directions are known is addressed. The approach taken here is to assume that the signal field at the array is comprised of P independent plane-wave arrivals from known directions. In practice, of course, the directions of arrival are rarely known exactly, however this difficulty can be overcome by using the standard MUSIC (Multiple Signal Classification) algorithm, which constitutes an angular pseudo-spectrum and an indicator of directions of arrival of different signals. The problem then reduces to estimating the signal powers from each of the P directions. Five estimators are presented for power estimation of acoustical sources. The conventional beamformer, the Capon estimator, the robust Capon estimator, the least squares fit estimator of the observed cross-spectral matrix of the array and a new estimator called the covariance vector estimator. Numerical and experimental results are presented showing that the performances of the covariance vector estimator are better than other estimators