ABSTRACT

Poor speech intelligibility in noisy environments is a common complaint of both users of conventional hearing aids and cochlear implants. A promising approach to this problem is the use of multiple microphones to reduce acoustic noise by exploiting the spatial information in the incoming acoustic signal. Multiple microphones can be mounted either in the same behind-the-ear hearing aid, in two hearing aid housings on either side of the head, or in more sophisticated, head-word arrays. The microphone signals can be combined by using fixed weighting factors, adaptive filters, or other, more elaborate algorithms. In contrast to single-channel noise reduction schemes which process one microphone signal only, substantial improvements in speech intelligibility have been reported for a wide range of acoustic environments and for several types of multi-microphone noise reduction systems. This paper reviews different multi-microphone noise reduction techniques, their performance in different acoustic environments, as well as evaluation procedures and performance metrics. Practical issues including placement of microphones, supporting algorithms, as well as current applications and research trends are discussed.