The Research On Speech Enhancement Of Digital Hearing Aids

The auditory sense of the patients with hearing impairment can be effectively improved by digital hearing aids,and it is an important way to improve the ability of language communication.In this thesis,the key algorithms of digital hearing aids are studied.The main work of this paper can be summarized as:Firstly,the principle of voice pronunciation,the principle of hearing and the characteristics of speech signal are described.Then this thesis discussed the key algorithms of digital hearing aids,such as sound source localization,speech enhancement,loudness compensation and acoustic feedback elimination.The strengths and weaknesses of these frequently used algorithms are also analyzed.Secondly,this thesis delved into the speech enhancement algorithm based on wavelet threshold denoising.Among the traditional threshold functions,hard threshold function and soft threshold function are most widely used.The hard threshold function is not continuous at the threshold,while there is a constant deviation between the original coefficient and restricted coefficient in soft threshold function.In response to these shortcomings,combine of these two kinds of threshold function,a new threshold function was proposed.At the same time,the traditional threshold setting rules are improved,and the threshold values of different layers are weighted.The results show that the new algorithm has better performance in speech enhancement.Finally,in-depth study of generalized sidelobe canceller(GSC).When there is an error in direction estimating,the target speech cannot be blocked by blocking matrix(BM)module completely.Then in the multiple-input canceller(MC)module,the target speech will be eliminated,which will cause the leakage of the target speech.In this thesis,a new optimization algorithm was proposed.First,we adjust the spectrum of the signal with time delay compensation,then the blocking matrix would be adjusted adaptively according to the characteristics of the correlation between the final output of MC module and the output of BM module.This way,the estimated direction can be closer to the real target speech direction in order to reduce the leakage of the target speech.