Hybrid Approach between LMS and Self-Correcting Adaptive Algorithm for Real Time Audio Noise Cancellation

Abstract

In present era of public address systems one of the most important issue is removing the noise signal from original signal. In various types of noise, which annoys audience, one of the most annoying problems is active feedback causing howling effect. This effect can be removed with the help of adaptive filters. There are various popular adaptive algorithms like LMS, NLMS, VLMS commonly used for this purpose. As public address system is a real time system, the delay in system should be minimum and for that the convergence time of adaptive filter is an important parameter. This paper presents comparison between LMS and self-corrective algorithm in SIMULINK and based on results, the hybrid approach is proposed which gives faster convergence time then parent algorithms. The proposed hybrid algorithm is implemented on FPGA platform altera cyclone II.