Algorithm Research For Digital Hearing Aids Based On Hearing Characteristics Of The Elderly

There are hundreds of millions of people with hearing loss around the world,and a large proportion of whom are elderly.The decline of speech recognition ability caused by hearing loss not only affects the verbal communication of the elderly,but also affects their physical and mental health.Using digital hearing aids can effectively improve the hearing of hearing loss patients,and significantly improve their speech recognition ability.This paper studied noise reduction algorithm in digital hearing aids and made some improvements.Introduced the auditory masking effect and auditory threshold characteristics of the elderly into the spectral subtraction noise reduction algorithm,in order to reduce the influence of music noise and other noises,and improve the subjective sense of hearing.The results of simulation analysis and subjective listening experiment show that the noise reduction algorithm can effectively reduce the intensity of noise and improve the SNR of speech.The introduction of masking effect designedly retained part of background noise,and it may reduce the SNR improvement brought by noise reduction,but can improve the subjective listening sense.This paper also designs a multi-channel loudness compensation algorithm based on longtime formant distribution.The cross frequencies of filter banks in the traditional multi-channel loudness compensation algorithm are fixed,and the formant structure may be destroyed when the formant frequency of the speech appears near the cross frequency.It would cause timbre distortion.This paper introduced the long-time formant distribution measurement into the traditional multi-channel loudness compensation algorithm.The statistical distribution of the formants frequencies is calculate with the algorithm from a long period of speech,and the cross frequencies of the filter banks are adjusted according to the statistical distribution of the formants.The results of simulation analysis and subjective listening experiment show that the improved multi-channel soundness compensation algorithm effectively reduces the damage of the formant structures caused by frequency division,and avoids the huge computation and the risk of timbre hopping caused by detecting the formant and switching cross frequencies of filter banks frame by frame.The effectiveness of the algorithms in this paper has been verified by objective simulation analysis and psychoacoustics subjective listening experiment.The results show that the improved algorithm can improve the subjective listening sense of speech under certain noise types and SNR conditions,and the results of variance analysis show significance.