| Digital hearing aid is an effective way for hearing impaired patients to improve their listening. Due to the close distance between the louderspeaker and the microphone in digital hearing aids, The acoustic feedback will easily occurs. The adaptive feedback cancellation algorithms can effectively eliminate the acoustic feedback signal and the quality of output voice from the hearing aid. This paper focused on the acoustic feedback cancellation algorithm on hearing aids and proposed two improved algorithms. The main work can be summaried as follows:Firstly, this thesis introduced the basic knowledge of voice, incentive model, channel model and radiation model. And then, compared the three basic methods of speech signal analysis, introduced some key technology of digital hearing aids and compared the performance of three basic algorithms, which are Least Mean Square algorithm, Recursive Least Square algorithm and Normalized LMS algorithm.Secondly, when a fixed step size is used in IPNLMS algorithm, the fast convergence rate and the low steady-state misalignment are conflict. To resolve this problem, this paper proposed a new variable step method instead of a fixed-step method. Compared with traditional fixed step size and variable step IPNLMS algorithms, the proposed algorithm can get faster convergence rate and lower steady-state misalignment.Thirdly, when NLMS algorithm and Proportion NLMS are used to process a speech signal, their performance will greatly decrease. In order to effectively eliminate acoustic feedback, this paper applied Affine Projection Algorithm to suppress the acoustic feedback signal in digital hearing aids. In order to improve the usefulness of the algorithm, this paper proposed a simplified, variable step size algorithm APA instead of a fixed step size one. Compared with fixed-step APA, the proposed algorithm can get faster convergence rate, lower steady-state misalignment and further reduce the amount of calculation. |
PDF Full Text Request