Research On Adaptive Feedback Cancellation In Ear-hanging Digital Hearing Aid

For people who are hard of hearing,digital hearing aids can effectively improve hearing level,and with the decrease of the volume of hearing aids,the phenomenon of echo has become increasingly serious,the presence of echo has seriously affected the comfort of the hearing aids and the intelligibility of speec h,the echo cancellation algorithm is more and more important in the speech processing,aiming at this problem,this paper conducted a study on the adaptive echo cancellation algorithm in ear-hanging hearing aid.This paper introduces the characteristics o f hearing and hearing impairment principle of ear,discusses the hardware structure and software algorithm of digital hearing aid in detail,the hardware structure includes a microphone,DSP processing module,a receiver,software algorithms includes speec h enhancement algorithm,loudness compensation algorithm,frequency compression algorithm,sound source localization algorithm,echo cancellation algorithm and so on.The principle of echo cancellation is introduced in the echo cancellation algorithm,which provides a basis for the research of adaptive echo cancellation algorithm in digital hearing aids.Based on the understanding of the principle of adaptive filter,the principle of Wiener filter and the steepest descent method is studied,analyzes the derivation process of Least Mean Square(LMS),Normal Least Mean Square(NLMS)and Recursive Least Square(RLS)algorithm.Step size is one of the key parameters of the algorithm,set LMS in different step parameters and simulate the performance,through the LMS,NLMS,RLS algorithm simulation,RLS algorithm has the fastest convergence speed,but the computation is large,and LMS algorithm is too simple,the performance is poor,NLMS algorithm has a moderate computation and with good performance,has been wi dely used in hearing aid.On the basis of the point-by-point updating algorithm,this paper introduces the algorithm of block processing,which greatly reduces the computational complexity,compares the performance of Block LMS(BLMS)and LMS,Normal BLMS(NBLMS)and NLMS algorithm,although the block processing algorithm of BLMS,NBLMS has low complexity,its performance is poorer.The NBLMS_M-K algorithm,which integrates the NBLMS and NLMS algorithms,is proposed,and it utilizes the idea of forgetting factor in RLS,where M is the size of the data processing cycle,K represents the first K coefficients of the adaptive filter.The simulations of the same K,different M and the same M,different K and different M,different K were compared.And for NBLMS and NBLMS_M-K are simulated and compared.The NBLMS,NLMS and NBLMS_M-K are analyzed synthetically,and we should select appropriate algorithms in the hearing aids according to the actual requirements of the system.