Research On Acoustic Echo Cancellation Algorithm Based On Adaptive Filtering

With the development of modern communication technologies,various communication devices have gradually become indispensable in our life and work.The quality of the call directly determines the most basic user experience.The main factors influencing the call quality are noise and echo.The echo is the sound of your own speech heard from the handset during a call which is often unbearable.How to reduce the influence of acoustic echo on the speech signal is the main research content of this article.Aiming at the shortcomings of the traditional echo cancellation algorithms,based on the principle of Acoustic Echo Cancellation(AEC),and the transverse finite impulse response filter and adaptive filtering algorithm,corresponding improved algorithm is proposed for AEC under different conditions,the main research content is as follows:(1)Under the condition that the input signal is irrelevant,the sparse control?-law proportional normalized least mean square algorithm(SC-MPNLMS)has high computational complexity for a complex and long echo path,a variable-step frequency-domain block SC-MPNLMS algorithm is proposed.Combining the logarithmic power spectrum(LPS)of the estimated echo and theerror signal,the regressive averaging method is used to process the correlation between the autocorrelation of each frequency point and the error signal to jointly determine the leakage coefficient,and the variable step matrix is used to replace the global step factor to update the filter weight vector,which improves the shortcomings of excessively large steady-state error caused by the excessively large step-length factor in the later algorithm.In order to solve the problem that the initial convergence speed of the frequency domain block adaptive filtering algorithm is slow,the initial weight of the filter is assigned to a random sequence with moderate sparse degree.(2)Under the condition of correlated input signals,and when there is impulse noise in the system,a jointly optimized normalized subband adaptive filter(NSAF)echo cancellation robustness algorithm is proposed.Based on the convergence of the standard NSAF algorithm,the algorithm is optimized by minimizing the mean square error(MSE)and combining the time-varying step size and the time-varying regularization parameter,so that the algorithm can achieve a low steady-state imbalance when obtaining a fast convergence rate,and when the impulsive noise occurs,the subband cut-off coefficient is updated in a recursive manner to improve the robustness of the algorithm to impulsive noise..(3)Also under the relevant input signal conditions,an improved sparse control proportional NSAF algorithm is proposed for the effect of echo path sparseness on the performance degradation of the NSAF algorithm.The sparsemeasurement method is used to estimate the sparseness of the unknown echo path,and the sparseness degree control equation is combined.The optimal parameters in the exponential equation are determined through simulation experiments.The piecewise linear function is used instead of the original exponential control equation to simplify the algorithm complexity.By estimating the sparseness of the echo path in each iteration,the robustness of the proportional NSAF algorithm to the sparseness of the echo path is improved.