Application of adaptive noise cancelling in a noisy reverberant environment

This paper describes a study that has been conducted on potential ways of estimating a signal coming from a specific source in a noisy reverberant room. That is, the desired signal is corrupted by reverberation and also signals coming from other sources, which may have similar spectral or statistical characteristics to the desired signal. By using an array of microphones and minimizing the effects of the undesired sources with the LMS algorithm, some significant improvements can be made. The results of some computer simulations will be presented, and their application to "in-situ" acoustic noise measurements will be discussed.