Reserch On Key Technologies Of Echo Cancellation And Loudness Compensation In Digital Hearing Aids

Digital hearing aids can effectively ameliorate the aural comprehension of the patients. However,the problem about echo cancellation and loudness compensation has restricted the performanceimprovement of hearing aid for a long time. In this paper, a deep research on basic principles of thetwo algorithms is carried on. We not only implement the existing algorithms, but also propose theimproved algorithms. The main work of this paper can be shown as follows:First, we researched the basic principle of digital hearing aid and the speech generatingmechanismn. The advantages and disadvantages of the speech processing algorithm which includes:acoustic sourch localization, speech enhancement, echo cancellation and loudness compensation arealso summarized.Second, we made a deep research on the adaptive echo cancellation algorithm: LMS, NLMSand RLS. To solve the slower convergence rate of NLMS and higer computation of RLS algorithm,an improved NLMS algorithm based on the discrete cosine transform is proposed. Experimentsshow that the proposed algorithm has a faster convergence than NLMS. At the same time, theproposed algorithm retained the high echo cancellation capability.Finally, to solve the problem that conventional multi-channel loudness compensation algorithmignores the protection of speech feature, a single-channel compression and amplification algorithmbased on multi-resolution wavelet is proposed. Multi-resolution wavelet is involved to decomposeand reconstruct, that can extract the envelope of signal spectrum. Then, calculate the required gainof each feature point. After that, the gain of entire signal spectrum is obtained by interpolationalgorithm. Experiments show that the proposed algorithm can not only compensate the loudness ofvoice, but also effectively protect the speech feature, which improves the speech recognition rate.