The Improvement of Audio Noise Reduction System Based on LMS Algorithm

This paper, based on the original LMS algorithm, proposed a novel method with wavelet preprocessing and signal is used as reference input to improve the noise reduction effect of the adaptive noise cancellation system. The new algorithm, in the presence of impulse noise, low SNR environment, has a higher amount of noise reduction than the original LMS algorithm based on adaptive noise cancellation system with better effect in reducing noise. The method first used wavelet packet threshold noise reduction algorithms, including impulse detection and de-noising, to remove the impulse part of the noise. Then the residual noisy signal is eliminated by adaptive noise cancellation system based on LMS which has signal as the reference input. Finally, the multi-group simulations are compared. Simulation results show that the improved adaptive noise cancellation system do have better noise reduction and practical value.