A second-order Volterra filter-based nonlinear clipping detector

Abstract

In this paper, we propose a novel nonlinear clipping detector based on the second-order Volterra filter for an acoustic echo canceller (AEC). Since the performance of the conventional AEC algorithm drastically deteriorates when nonlinear clipping occurs, the adaptive filter pauses adaptation during the nonlinear clipping periods to avoid unwanted divergence. These nonlinear clipping periods are detected in our method using the magnitude spectrum of the quadratic Volterra filter in lower frequency ranges. Through extensive computer-based simulation results considering an acoustic room environment and various clipping levels, it is found that the proposed approach is effective in detecting the nonlinear clipping periods and improving AEC performance compared to the conventional AEC algorithm.