Research On Multichannel Echo Cancellation Algorithm For The Audio And Video Conference System

In recent years,with the development of material technology,information processing technology and communication technology,the audio and video conferencing system has overcome the limitations of spatial and geographical location,and gives people a method of long-distance interactive communication.The audio and video conference system allows people to make voice calls or video chats at any time,reducing the cost of communication and improving office efficiency.In order to improve the call quality of the audio and video conference system,an adaptive filter is used to eliminate the influence of the echo signal on the voice call.However,as people have higher demands on the call performance of the audio and video conference system,the audio and video conference system will be equipped with multiple microphone devices and speaker devices.The existing echo cancellation method cannot be directly applied to such scenarios,so it needs to be analyzed and designed an appropriate multichannel echo cancellation algorithm.Multichannel echo cancellation scenes can be divided into multimicrophone echo cancellation scenes and stereo echo cancellation scenes according to the number of speaker devices.In multimicrophone echo cancellation,the adaptive filter algorithm have the problems of slow convergence speed,high calculation amount,and poor echo cancellation effect.In stereo echo cancellation,the adaptive filter algorithm have the problem of large misalignment.Based on the existing single-channel echo cancellation,the following aspects were studied in this paper:(1)Aiming at the problem that the traditional time domain filter will slow down the convergence rate of the filter when processing speech signals in multimicrophone echo cancellation,a subband filter system with a multiband structure is studied,and this system is extended to the multimicrophone echo cancellation problem.This multimicrophone echo cancellation system can ensure that when the input signal is a speech signal,the echo cancellation has a faster convergence speed.Simulation experiments show that this method has a faster convergence speed when processing speech signals.(2)Aiming at the problem that when multiple adaptive filters estimate multiple echo signals,the amount of calculation of the system will increase in multimicrophone echo cancellation,the relative transfer function is used to estimate the echo signals collected by other microphones.The RAP algorithm is also used to further improve the effect of using relative transfer functions to estimate echo signals.Considering the problem that the calculation amount of this algorithm will increase when the number of iteration is high,the fast form of the RAP algorithm is derived to reduce the calculation amount of the improved algorithm.This fast algorithm ensures that the system has a low amount of calculation,and at the same time it can get a better echo cancellation effect.Simulation experiments show that the fast algorithm can get a better echo cancellation effect.(3)Aiming at the problem that the traditional nonlinear two-way NLMS algorithm has a large amount of misalignment in stereo echo cancellation,the RAP algorithm is used to improve the effect of the two-way NLMS algorithm for stereo echo cancellation.Considering the problem that the calculation amount of the improved algorithm will increase when the number of iteration is high,a fast form of RAP algorithm is proposed to reduce the calculation amount of the improved algorithm.This fast algorithm ensures that the system has a low amount of calculation,and at the same time can obtain a low amount of misalignment.Simulation experiments show that the fast algorithm can reduce the misalignment of the two-way NLMS algorithm.