Research On Adaptive Echo Cancellation And Noise Cancellation

It is inevitable that echo and noise can exist in communication systems; with the continuedevelopment of video conference and hand free system, the requirement of AEC has beenemphasized for efficient algorithm can meet the requirement for the real-time system and highquality. As the echo is time varying, and the adaptive technology is widely applied into theecho cancellation for adaptive filter has the characteristic of tracing the changing input signals.The influence of Interference signal is serious on AEC and should be removed. It is commonto face some problems such as acoustic echo, noise and doubletalk in building up acommunication system environment. To achieve preferable communication quality withoutinterference effect, an integrated acoustic and noise canceller is adopted and two coretechnologies are discussed in this paper: Remove the nonlinear distortion caused by electroniccomponents, and analyzes the algorithm of adaptive process control; the protection algorithmduring double talk environment.In this paper, the following major work was done:(1) The background knowledge of echo cancellation is introduced, includes its meaning,the current situation and the developing trend at home and abroad, and introduced thetechnical difficulties of adaptive echo and noise cancellation.(2) The principle of adaptive echo cancel is introduced, the algorithm description,convergence and the applicability in adaptive echo offset are discussed of commonlyused linear echo cancellation algorithm such as least mean-square error (LMS),normalized least mean-square error (NLMS), recursive least squares (RLS) and affineprojection algorithm (AP), respectively. Finally compared these algorithms with thetechnical criteria of echo cancellation.(3) It is maybe that the traditional AEC lose cancel effect, for nonlinear distortionscaused by amplifier and speaker. So in this paper we introduced an identificationmethod using the pseudo-coherent function, and adopted an echo cancellationstructure suitable for memory less nonlinear environment, analyses identificationnonlinear coefficient and adaptive process control algorithm respectively. On the baseof several variable step size LMS algorithms, proposed a novel variable step sizealgorithm which is based on sigmoid function, the suitable for using in Adaptive echo cancellation of this new algorithm is incarnate by computer simulations.(4) It is common that double talk situation occur during talking circumstance, so in thispaper, we introduced the development of double talk detection and gave sometechnical difficulties and defects, and then we employed an algorithm without usingDTD but can distinguish DT from EPC and with good double talk protectionperformance.(5) Considering the effect of noise in echo cancellation, so in this paper mainly use anoise cancel method which is based on the optimal wiener filter, and combined it withthe AEC got a combination structure with the function of adaptive noise and echocancellation. To proving the effectiveness of the structure, using noise and actualspeech as input signal during the computer simulation, respectively.