A fast converging and stable adaptive filtering algorithm for stereophonic acoustic echo cancellation

A new automatic switching algorithm (ASA) has been introduced in this paper for use in stereophonic acoustic echo cancellation (SAEC) in order to achieve fast convergence rate with low computational cost whilst guaranteeing algorithm stability. This algorithm has been designed to overcome the problem of "abrupt changes in the acoustic echo paths" (ACEP) and "double-talk" (DT). Two different detectors, one based upon the estimate of the autocorrelation between the error signal e(n) and e(n — 1) and another based upon the energy ratio of the desired signals d(n) at different conditions, are employed within the ASA to detect ACEP and DT respectively. Simulation results are presented based on real stereophonic signals to support these ideas.