Multichannel Acoustic Echo Cancelation in Multiparty Spatial Audio Conferencing with Constrained Kalman Filtering

This paper proposes a novel scheme for acoustic echo cancellation (AEC) in multi-party, spatialized audio conferencing by exploring the constraints among the adaptive filters for each synthesized remote speaker. The AEC algorithm employs constrained Kalman filtering (CKF-AEC) that takes advantage of the uncorrelated reference signal from each remote channel [1], yet restricts the filter adaptation within the subspace determined by the spatialization functions used to virtualize each remote participant and the number of unique acoustic paths between the loudspeakers and microphone. In addition, the proposed algorithm allows the adaptation for channels without input signals. Experimental results show the proposed algorithm has much higher performance than the scheme proposed in [1] which uses NLMS as the adaptive filter and does not explore the available constraint.