DTD-free nonlinear acoustic echo cancellation based on independent component analysis

An adaptive acoustic echo cancellation method based on independent component analysis, which can remove nonlinear echo components, is presented. Since this method takes into account the independence between a far-end signal and the echo-cancelled output to cancel acoustic echoes which are independent of near-end signals, it can continue to update AEC parameters even in double-talk situations, thereby improving the cancellation performance. Experiments show that the proposed method provides better performance than the conventional least-mean-square methods for both linear and nonlinear AEC systems without requiring separate double-talk detection (DTD).