Wave-domain echo-path model with aliasing for echo cancellation

Abstract

Wave-domain adaptive filtering for echo cancellation has been proposed for achieving immersive full-duplex sound conferencing that uses wave field reconstruction as spatial sound rendering. In wave-domain adaptive filtering, fundamental solutions of the wave equation are spatially sampled and used as the orthogonal basis functions. This sampling is determined by loudspeaker spacing and results in aliasing; aliasing occurs above a few thousand Hz for spacing of several centimeters. The goal of this work is to investigate the effect of applying adaptive filtering on echo signal with aliasing when the loudspeaker array and microphone array are uniform linear arrays of identical geometries. We came to the conclusion that we can apply the wave-domain echo-path model, used below spatial Nyquist frequency, to wave-domain adaptive filtering over this frequency even in the presence of aliasing components.