Adaptive Noise Cancelling for audio signals using Least Mean Square algorithm

The methods to controlling the noise in a signal have attracted many researchers over past few years. One such approach is Adaptive Noise Cancellation which has been proposed to reduce steady state additive noise. This method uses two inputs - primary and reference. The primary input receives signal from the signal source which has been corrupted with a noise uncorrelated to the signal. The reference input receives noise signal uncorrelated with the signal but correlated in some way to the noise signal in primary input. The reference input is adaptively filtered to obtain a close estimate of primary input noise which is then subtracted from the corrupted signal at the primary input to produce an estimate of a clean uncorrupted signal, which is the Adaptive Noise Cancellation output. A desired signal corrupted by noise can be recovered by feeding back this output to Adaptive Filter and implementing Least Mean Square algorithm to minimize output power. The audio signal corrupted with noise is used as a primary input and a noise signal is used as reference input. Computer simulations are carried out using MATLAB and experimental results are presented that illustrate the usefulness of Adaptive Noise Cancelling Technique.