Nonlinear acoustic echo cancellation using a Hammerstein model

摘要: In hands-free telephone or video conference application, there exists an acoustic feedback coupling between the loudspeaker and microphone, which creates echo. Linear echo cancellers (AECs) are commonly used to remove this However, they unable effectively cancel nonlinear distortions. This paper employs a Hammerstein model describe channel of system concatenated with linear faded path. A feed-forward neural network is static nonlinearity finite impulse response (FIR) structure dynamic system. The formed applied real data collected in anechoic chamber it performs slightly better than models. Although improvement small, results show some interesting insights on characteristic loudspeaker's nonlinearities their effect performance AEC.