Research On Key Technologies Based On Wavelet Transform In Digital Hearing Aids

Wearing digital hearing aids improves the ability of hearing impaired patients to adapt to different sound environments.Speech enhancement and loudness compensation are the key technologies of digital hearing aids.Based on the study of traditional algorithms of speech enhancement and loudness compensation,this paper proposes an improved algorithm.The main work is as follows:Firstly,this paper analyzes the core idea of digital hearing aids,and studies the theory of speech signal that includes the generating principle,characteristic and auditory principle.Three mathematical models and analytical methods are mentioned.The core technology of digital hearing aids is studied.This paper focuses on the speech enhancement and loudness compensation algorithm.Secondly,the problem of speech leakage and residual noise caused by generalized sidelobe canceller adaptive beamforming algorithm is studied.Based on the analysis of the causes of speech leakage and the introduction of wavelet threshold denoising technique,an adaptive beamforming algorithm is proposed.Wavelet thresholding can improve the SNR of the speech signal effectively,which makes the speech singal more accurate after the GSC structure.Simulation results illustrate that the algorithm reduces speech leakage and residual noise in a way,and improves the speech recognition.Finally,the algorithm of loudness compensation based on wavelet transform is proposed.The existing wavelet domain loudness compensation algorithm only compensates the loudness of the low frequency,because the high frequency part carries less information and the high gain would make the hearing loss patients suffer secondary damage.In this paper,we combine wavelet domain and frequency shift technology to achieve loudness compensation algorithm,which transfers high frequency signal to low frequency and then loudness compensation is reached.Results show that the algorithm protects the integrity of speech signal,thereby improving the comfort and recognition rate of speech singal.