Room Frequency Response Equalization Using Modified X-Filtered Adaptive Algorithms

Abstract

This article presents a technology of the acoustic frequency response equalization. It is based on the usage of the modified x-filtered adaptive algorithms in the equalizers. Comparing to the traditional x-filtered algorithms, the modified ones allow to accelerate the equalizer convergence if the computationally simple LMS or the NLMS adaptive algorithms based on the gradient search strategy are used for the equalizer weights computation. Besides, the modification allows to use the computationally complex but simultaneously efficient RLS adaptive algorithms in the x-filtered mode because the algorithms cannot be used in the traditional x-filtered mode. The proposed equalizer architectures and the computational procedures of the traditional and modified adaptive filtering algorithms are presented. Simulation demonstrates a better performance of the modified algorithms comparing to the traditional ones in the following terms: the equalized frequency response ripple value, the proximity of the PSD curves of the undistorted speech signal and the signal passed through the equalizer and acoustic medium, the initial duration of the transient response of the error signal and the response steady-state value.