Abstract

This project uses the LMS algorithm for reducing the noise of a speech signal that has been corrupted with noise. The problem is done for a car where the speech signal is being corrupted by the noise of the engine. The noise from either the rare of the car or the engine itself can be used to reduce the noise from the corrupted speech signal. In the following first a brief introduction of digital signal processing is given follwed by a short summary of adaptive filters. The proof for the condition that the reference noise and the original speech siganl are correlated is not given here but the refernce has been pointed. Then the use of the LMS algorithm is given followed by my system configuration, experiments and results. Finally I conclude by saying what I think about the work I have done.


gsm@ee.duke.edu
Fri Mar 08 18:41:27 1996