Research And Application Of The Enhancement Method For Initial Of Chinese Syllable

Senile hearing loss is a high-morbidity disease that seriously harms the physical and mental health of the elderly,and wearing hearing aids is one of the most effective ways to compensate for senile hearing loss.The existing researches on hearing aid algorithms for elderly patients with hearing loss whose native language is Chinese are based on some parameters in the time domain or frequency domain.They focus on achieving digital hearing aids through traditional or deep learning methods,which ignore the unique acoustic properties of initials and finals in Chinese speech.This thesis studies the problem of digital hearing aids for elderly patients with hearing loss whose native language is Chinese.This thesis introduces the acoustic characteristics of Chinese phonetic initials and finals and mature speech recognition technology into the traditional hearing aid algorithms and a digital hearing aid algorithm based on the initial enhancement of Chinese Mandarin speech syllables is proposed.The main work of this thesis includes:(1)This thesis studies the acoustic characteristics of the initials and finals in Mandarin and the auditory characteristics of elderly patients with hearing loss.A prior database of the initials to be enhanced for typical elderly patients with hearing loss is constructed,including the type and the gain value of initials to be enhanced in the initial enhancement algorithm.(2)Mandarin Chinese is sampled and converted into a digital speech stream,and then the digital speech stream is converted into Chinese character text using mature speech recognition technology.Referring to the first-level and second-level Chinese characters and their pinyin specified in "GB2312-80" to find the pinyin corresponding to a Chinese character to convert the Chinese character text into a continuous pinyin stream.(3)The existing syllable segmentation algorithms are studied in this thesis.A two-parameter and two-threshold syllable segmentation algorithm based on the Teager energy after shortterm empirical modal decomposition and Mel frequency cepstral distance is used to segment the syllables in the time domain to form discrete pinyin streams.(4)An algorithm of initials and finals segmentation based on the typical finals formants and binary spectral matrix is proposed.Each syllable in the discrete pinyin stream is screened according to the established prior library of initials to be enhanced.For the syllables that need to be enhanced,the initials and finals of the syllables are firstly segmented according to the initials and finals segmentation algorithm proposed in this thesis,and then the initial need to be enhanced according to the degree of hearing loss of the initial.Finally,the syllable should be synthesized with the enhanced initial and the original final and outputted.In order to verify the correctness and effectiveness of the algorithm proposed in this thesis,a series of simulation experiments are designed to verify the proposed initial enhancement algorithm.The results show that the specific initials can be enhanced by the Chinese syllable initial enhancement algorithm in the daily laboratory language environment,so as to compenste for he hearing loss of some initials in elderly patients with hearing loss.