An adaptive LMS virtual microphone

It has been shown recently that, using two (or more) microphones and a forward-difference approximation, it is possible estimate the pressure at a remote observer location rather than at the actual microphone locations. These “virtual-microphones” have been successfully used in active noise control systems to minimise the pressure at a location remote from the microphones. However, the virtual microphones have not been without their problems. Determining the open loop weights for the microphone elements is not always a trivial task. In practice the implementation of virtual microphone arrays has additional difficulties. The virtual sensors have shown that they are sensitive to phase and magnitude mismatches between microphone elements, so it is essential that well matched elements are used. This paper explores the use of the LMS algorithm to derive the optimal weights for the microphone elements. It will be seen that this optimal approach improves the quality of the pressure estimate at the desired remote location compared to the traditional open loop approach. It completely compensates for amplitude mismatches and spatial errors, and the method even addresses phases mismatches to some extent.