Objective Assessment Of Hearing Aid Performance Based On Gaussian Process

As noise pollution and the aging of the population become more severe,the number of people with hearing impairment is increasing.In order to improve the level of hearing,the role of hearing aids is prominent,and its voice quality is directly related to the health of hearing impaired patients.Therefore,it is important to objectively evaluate the hearing aid’s voice quality accurately.In order to obtain real and reliable medical data,the research process of the dissertation was combined with China Academy of Metrology and Peking Union Medical College Hospital through a large number of simulation experiments to realize the experimental process of combining theory with practice.This subject proposes an objective evaluation method of speech quality based on Gaussian process,and improves it based on this algorithm;designs a cluster analysis model,introduces an asymmetric kernel function based on the original Gaussian process,and designs a hearing aid based on Python language A platform for objective evaluation of speech quality.Simulation experiments have proved that the Gaussian process method and improved algorithm proposed in this paper can evaluate the speech quality of hearing aids,and the improved algorithm is more accurate.The main research work and innovations of the thesis are as follows:(1)A test environment for a hearing aid objective evaluation system is set up,and a voice database for hearing aids is established based on the test platform to collect voice signals and process them.The characteristic parameters of speech signals were studied in the time and frequency domains,and the characteristic parameters of hearing aids were designed and implemented.(2)A Gaussian process method was proposed to achieve objective evaluation of hearing aids.This method first optimizes the features,then establishes an objective evaluation model of the hearing aid,and obtains an objective score through simulation experiments.The relevance and accuracy of objective scoring using Pearson’s correlation coefficient are described.The results show that the algorithm is effective.(3)Based on the original Gaussian process,an improved method of multi-kernel covariance function is proposed,and a suitable multi-kernel covariance function is selected to improve the original model.The clustering algorithm is added to the experimental process to cluster the experimental data to improve the prediction accuracy.By analyzing the experimental simulation results,the algorithm performance evaluation indexes are summarized,and the improved algorithm is verified to be effective and feasible with high accuracy.(4)On the basis of cluster analysis,an objective evaluation method of hearing aid speech quality based on Gaussian process method is proposed—optimizing the asymmetric kernel function method of lengthscale Li in kernel function in Gaussian process.The algorithm first clusters the voice data into different categories,and then uses different kernel function parameters to evaluate the Gaussian process objectively for different categories of data.Through simulation experiments and algorithm evaluation indicators,the improved algorithm is verified,and it is confirmed that the algorithm can accurately evaluate the speech quality of hearing aids.