Acoustic Echo Cancellation using the DUET Algorithm Based Blind Separation in a Noisy Environment

Abstract

In hands-free full-duplex communication systems, acoustic noises picked up by the microphones are mixed with echo signals, which may result in poor echo cancellation performance of the communication system. In such cases, a degenerate unmixing estimation technique, adjusted in the time-frequency domain, can be employed to separate undesired echo signals and noises. However, scaling and permutation ambiguities still have to be solved for blind source separation (BSS). In this paper, a new acoustic echo cancellation approach, utilizing the DUET algorithm based BSS, is proposed, whereby a new scale estimation method is introduced along with kurtosis-based desired signal selection. Simulation results demonstrate that the proposed approach yields better echo cancellation and noise reduction performance, compared with conventional methods.