Abstract

In this paper, a new adaptive filtering algorithm for Stereophonic Acoustic Echo Cancellation (SAEC) is proposed in order to obtain computational complexity reduction and convergence rate improvement. The proposed algorithm is based on a hybrid mono/stereo structure between Fast Affine Projection (FAP) and Frequency-domain Normalized Least Mean Square (F-NLMS) algorithms, together with subband processing. Simulation results, based on recorded speech signals, suggest that convergence rate can be improved at low computational complexity.