Stereo echo cancellation algorithm using adaptive update on the basis of enhanced input-signal vector

Stereo echo cancellation requires a fast converging adaptive algorithm because the stereo input signals are highly cross correlated and the convergence rate of the misalignment is slow even after preprocessing for unique identification of stereo echo paths. To speed up the convergence, we propose enhancing the contribution of the decorrelated components in the preprocessed input-signal vector to adaptive updates. The adaptive filter coefficients are updated on the basis of either a single or multiple past enhanced input-signal vectors. For a single-vector update, we show how this enhancement improves the convergence rate by analyzing the behavior of the filter coefficient error in the mean. For a two-past-vector update, simulation showed that the proposed enhancement leads to a faster decrease in misalignment than the corresponding conventional second-order affine projection algorithm while computational complexities are almost the same.