Optimal power allocation for speech enhancement using remote wireless microphones

Abstract

Speech capture in the presence of non-stationary interfering signals, e.g. music or other competing talkers, is a challenging problem, and in a recent work, distributed microphones have been employed for this purpose. In this reported work, consideration is given to the problem of power allocation in a distributed microphone system, where a directional wireless microphone is placed close to each interferer and provides an estimate of the statistics of the undesired signal for suppression at the primary microphone. The optimal transmit power allocation is derived such that the mean squared error in estimating the desired signal is minimised subject to a total power constraint. It is shown that signal-to-noise performance improves at the primary microphone using optimum power allocation in comparison with an equal power allocation scheme. © The Institution of Engineering and Technology 2013.