Publication
Acoustic echo control and noise suppression is an important part of any "handsfree" telecommunication system, such as telephony or audio or video conferencing systems. Bandwidth and computational complexity constraints have prevented that stereo or multi-channel telecommunication systems have been widely applied. The advantages are very low complexity, high robustness, scalability to multi-channel audio without a need for loudspeaker signal distortion, and efficient integration of echo and noise control in the same algorithm.; The proposed method of processing multi-channels audio loudspeakers signals and at least one microphone signal, comprises the steps of: transforming the input microphone signals (y 1 (n), y 2 (n), . . . , yM(n)) into input microphone short-time spectra, computing a combined loudspeaker signal short-time spectrum [X(i,k)] from the loudspeaker signals, (x 1 (n), x 2 (n), . . . , xL(n)), computing a combined microphone signal short-time spectrum [Y(i,k)] from the input microphone signal, (y 1 (n), y 2 (n), . . .; , yM(n)), estimating a magnitude or power spectrum of the echo in the combined microphone signal short-time spectrum, computing a gain filter (G(i,k)) for magnitude modification of the input microphone short-time spectra, applying the gain filter to at least one of the input microphone spectra, converting the filtered input microphone spectra into the time domain (e 1 (n), e 2 (n), . . . , eM(n)).