| With the rapid growth of population aging, the population of hearing loss is increasing year by year. In modern medicine, the method of treatment is limited for hearing loss, so wearing a hearing aid becomes the best solution at present. Because of small volume, low power consumption and powerful function of digital hearing aids, it occupies the main position of the hearing aids market at the present stage. Loudness compensation method is one of speech processing algorithms for digital hearing aids, which can carry on nonlinear compensation for speech signal, enhance the hearing level of patients. The paper mainly studies the loudness compensation algorithms for digital hearing aids, the main research works of this paper is as follows:This text introduces the research background, development history of hearing aids, focuses on the research status of hearing aid technology and loudness compensation technology. The paper analyses the mechanism of speech signal and the characteristics of hearing loss according to audiology, which provides a theoretical basis for loudness compensation algorithms.In-depth study of wide dynamic range compression(WDRC) algorithm and eight channel loudness compensation algorithm which is based on human auditory characteristics. On the basis of these algorithms, in the light of the inaccuracy of feature points is extracted by wide dynamic range compression algorithm, a WDRC algorithm according to the resonance peak estimation is proposed. However, medical studies have shown that when hearing loss is more than 60 d B, high gain for high frequency voice will not only fail to compensate for the purpose, but also have an adverse effect on speech understanding. In order to solve this problem, the paper introduces compression and transfer frequency technology, to ignored the protection of the voice resonance peak by multi-channel loudness compensation algorithm, combine multi-resolution analysis technique based on wavelet, a wavelet-single channel loudness compensation algorithm based on frequency transition is put forward.The detailed simulation verification and comparison for each loudness compensation algorithm are carried out by using MATLAB. The simulation results show that the eight channel loudness compensation algorithm and wide dynamic range compression algorithm have caused serious distortion of speech. For the wide dynamic range compression algorithm based on the resonance peak estimation, which can effectively reduce the nonlinear distortion of speech resonance peak, but its compensation amplitude is relatively small.For the wavelet-single channel loudness compensation algorithm based on frequency transition, which can improve the hearing level, at the same time, effectively enhance the speech recognition rate of patients, to ensure auditory comfort degree of patients. |
PDF Full Text Request