Broadband active sound quality control based on variable step size filtered-x normalized least mean square algorithm

Abstract

Active noise control (ANC) provides a suitable solution for low frequency noise control. The aim of conventional ANC is to attenuate the unwanted noise as much as possible without consideration of its frequency components. However, it is desirable to retain certain sounds with specified spectral content to improve the human perception of sound. Active noise equalizer (ANE) is a method proposed to solve this problem. Traditional ANE uses standard filtered-x least mean square (FXLMS) algorithm. To accelerate the convergence of the controller while retaining low mean square error (MSE), we propose to use the variable step size filtered-x normalized least mean square (VSS-FXNLMS) algorithm to replace the FXLMS algorithm when updating the controller. The adaptation step size of this algorithm is adjusted dynamically according to the input signal and the instantaneous pseudo error of the controller. Numerical simulation results demonstrate that VSS-FXNLMS algorithm exhibits rapid convergence with low MSE.