Geometric Research On Speech Enhancement Algorithms For Digital Hearing Aids

With the development of semiconductor technology, high-end digital hearing aid has a powerful digital signal processing ability today, especially we enter the digital era in the recent ten years, digital hearing aid becomes the hope of listening patients with its low noise, energy saving, miniaturization and easy to operate features. The development of digital hearing aid mainly includes two aspects, algorithm research and shape design. The former problem is how to design algorithms can make patients hearing comfortable, while improving the sensitivity of speech. These algorithms include speech enhancement, dynamic range compression, frequency shift technology and automatic gain control. The shape should design a beautiful appearance, wearing comfortab le and reduce the weight and volume as far as possible. The above two aspects are the important factors that influence the development of digital hearing aid, also they are the research direction of future hearing aid.The main research direction of this thesis is speech enhancement algorithm for digital hearing aids, based on the depth understanding of the traditional speech enhancement algorithm, this paper designs a Mel frequency speech enhancement algorithm based on Wiener filter algorithm, and designs each part of the processing module to realize speech enhancement function.The paper presents the algorithm of voice activity detection(VAD) with a small amount of calculation and high accuracy. Only with a high effective VAD algorithm can we improve the effect of the speech enhancement algorithm.In order to improve the signal-to-noise ratio of the output speech, this paper designs a gain factor based on the priori signal-to-noise ratio. The purpose is to reduce the residual noise while ensure the quality of speech, so that the output speech SNR can be further improved.After the algorithm is designed, this paper has a full verification of different algorithms, analyzes the advantages and disadvantages of each algorithm. The main problem of traditional algorithm is the low quality of speech processing, speech loss is severe, producing music noise significantly, and the computation is not suitable for digital hearing aid. The experimental results show that, the speech enhancement algorithm designed in this paper has a high speech quality, at the same time, the SNR is significantly higher than other kinds of algorithm. In the computation calculation, relative to other complex algorithms, using the enhanced algorithm designed in this paper is more suitable for digital hearing aid.Finally, the algorithm is successful in the realization of digital hearing aids development based on TMS320VC5509. Design the data flow and software with a detailed analysis of the basic operation and make a fixed point processing, discusses the problems in the procedure and optimize the program. The algorithm processing effect is good with simulator validation.