Acoustic Echo Cancellation In A Two-way Call Scenari

With the rapid development of science and technology,the communication between people has overcome geographical barriers.The popularization of electronic communication equipment reduces the cost of communication,so that people can make voice calls anytime and anywhere,and communication becomes more convenient and efficient.However,the existence of acoustic echo seriously affects the quality of voice calls.Especially in the double-talk scenario,both parties not only hear each other’s voice from the communication device,but also hear their own voice,which brings a bad communication experience to both sides of the call.At present,the mainstream method of acoustic echo cancellation is to use adaptive filtering algorithm.Based on this,this paper mainly studies the application of adaptive filtering method to achieve acoustic echo cancellation in the scenario of double-talk.The main research contents are as follows:(1)Aiming at the problem that the traditional adaptive filtering algorithm in the linear acoustic echo cancellation system has limited performance in the double-talk scenario and the poor performance in the environment with high reverberation degree,the Kalman algorithm and the Normalized Least Mean Square algorithm(NLMS)are combined to realize the KN method of acoustic echo cancellation by using the double filter structure.Taking advantage of the fact that the Kalman algorithm can work normally in the scenario of double-talk,the KN method uses a double filter structure to update the two-filter by using the energy of the error signal output by the Kalman filter and the NLMS filter as the basis for the echo suppression ability.The simulation results show that the KN algorithm has better echo cancellation performance in the double-talk scenario with high reverberation.(2)Aiming at the balance between convergence speed and stability of adaptive filter algorithm based on gradient descent in double-talk scenarios,and the performance degradation of Double-Talk Detector(DTD)and adaptive step strategy in continuous double-talk scenarios,a Recursive Least Square(RLS)variant algorithm based on semi-blind source separation structure is proposed.This method does not require the use of DTD,and the acoustic echo cancellation is realized in the frequency domain by using independent component analysis based on an auxiliary function,and the auxiliary function technique is used on the basis of fullduplex characteristics to avoid explicit step parameter selection.Simulation results show that the proposed method has lower computational complexity and better echo cancellation performance in the double-talk scenario.