Research On The Active Noise Control In Digital Hearing Aids

In recent years, with the development of digital signal processing, digital hearing aids have made a great contribution to the hearing rehabilitation of many hearing disabilities. Nowadays, most of the hearing aids are relatively small, and the microphone is close to the headphone, so it is easy to cause echo or howling, which is complained by most of the hearing aids users. Therefore, in this thesis, the active noise control method is taken to reduce the noise in digital hearing aids, so as to improve the ability of the patient’s language recognition. The study of the subject has high academic value and potential applications prospect.Firstly, a detail introduction to the adaptive filter algorithm and active noise control theory is made, especially for the Newton’s method and the steepest descent method in detail. Based on this, the minimum mean square error (LMS) algorithm and normalized minimum mean square error (NLMS) algorithm are also introduced.Then the feedforward and feedback active noise control are presented. And based on the minimum mean square error algorithm, the feedforward filtered minimum mean square error algorithm (FX-LMS) is researched. Firstly an estimated recognition to the sub-channel is made, and then the noise is controlled using the minimum mean square error algorithm. For the different types of noise signal with low frequency, high frequency or white Gaussian noise, the study finds that a better noise-canceling effect can be obtained for the noise with low frequency.According to the knowledge the howling of the hearing aids is mainly caused by the frequency component higher than2.2kHz for the hearing impaired. Therefore, this thesis makes an improvement on the NLMS algorithm and proposes the band-limited NLMS algorithm. A simulation study is made to the finite impulse response and infinite impulse response adaptive filter, respectively, which indicates that band-limited NLMS algorithm can eliminate the howling frequency component effectively, thereby further reduce the noise in the output speech signal. Meanwhile, the infinite adaptive filter can achieve the same denoising results with lower order that is to say with much simpler system structure.Finally, a hardware design for digital hearing aids in DSP board TMSVC5509A is introduced. The flow software chart is firstly described of the feedforward FX-LMS algorithm that is implemented on the DSP board. Then, some preliminary denoising results are obtained in the compiler environment CCS3.3.