Low-complexity implementation of a real-time decorrelation algorithm for stereophonic acoustic echo cancellation

Multichannel acoustic echo cancellers are used in teleconferencing systems in order to reduce undesired echoes and to introduce better performance in terms of sound localization in the presence of more than one participant. Already, stereophonic systems obtain realistic performance, since listeners have spatial information that helps to identify the speaker position. Unfortunately, the correlation between the two channels prevents a correct identification of the echo paths and signal decorrelation is needed. In this paper, a low-complexity implementation of a real-time decorrelation approach is proposed based on the “missing-fundamental” phenomenon. This implies that the proposed approach is focused on the computation of one adaptive parameter in order to guarantee high performance in terms of convergence speed, audio quality, and computational complexity. Experiments confirm the effectiveness of the proposed approach, showing its feasibility in real scenarios.

► An approach for channel decorrelation in stereophonic acoustic echo cancellation is proposed.
► The approach is completely structured around only one coefficient that varies with the fundamental frequency.
► A pre-emphasis stage is considered for a better fundamental frequency estimation and tracking.
► The real time implementation of the proposed approach on a PC has been carried out.