Use of intermicrophone correlation in estimating signal to noise ratio

Abstract

This thesis presents the design, analysis, and simulation of a system that uses the correlation coefficient of audio inputs gathered at two spatially separate microphones to determine the signal to noise ratio in the environment. This work is motivated by past research in microphone array hearing aids, where accurate estimates of SNR were shown to improve performance. Signal to noise ratio is defined as the ratio of energy in the direct component (audio sources originating in front of a broadside array) to energy in the interference component (sources originating from the sides of the array). The design presented is a simple hypothesis testing mechanism for determining whether the SNR exceeds a fixed level. In the analysis, behavior of the system is studied theoretically under varying conditions of reverberation in the environment, and processing parameters are determined to optimize system performance. Finally, simulations test the true performance of the system to verify the validity of the theoretical analysis.