Objective assessment of hearing aid performance based on Gaussian process
Xiaomei Chen, Liu Guodong, Zhong Bo · 2019
Hearing aid performance evaluation has always been a concerned researchers. Efficient and accurate objective evaluation of hearing aid performance is a ultimate goal. Due to the complexity low efficiency and weak operability of subjective assessment process. In this paper, an Gaussian process regression based algorithm is put forward for the objective evaluation of hearing aid performance. Firstly, the speech features are extracted for the hearing aid output, then, Gaussian process model with some improvement on kernels selection is built to map the features to subjective scores. Finally, the objective scores obtained by Gaussian process model is compared with the subjective scores. And PCC parameter is used to measure the consistency and accuracy. Meanwhile the performance Gaussian process model is compared with the multivariate adaptive regression spline method(MARS) under the same data, result shows that the algorithm in this paper is advantageous than MARS.