关键词: 信息处理技术, 高斯混合模型, 遗传算法, 模糊最小目标函数准则, 说话人识别

Abstract: The traditional training method of Gaussian mixture model (GMM) is sensitive to the initial model parameters and easy to lead to a suboptimal model in practice. To solve this problem, a hybrid algorithm was proposed to optimize GMM model parameters globally and to improve parameter precision by using genetic algorithms (GA) and fuzzy clustering estimation method. An adaptive mixture crossover rate and mutation rate of GA and fuzzy minimum objective function (FMOF) algorithm were used to form a hybrid operation for reestimation of the model parameters, which can improve GA convergence speed. Experiment results based on PKUSRSC database show that this algorithm can obtain more optimized GMM and better than traditional GMM training method and maximum simulation training methods.

Key words: information processing, Gaussian mixture model, genetic algorithms, fuzzy minimum objection function(FMOF) rule, speaker recognition