中文摘要：

传统的最小二乘支持向量机（LS—SVM）使用特征向量作为训练样本，在说话人识别系统中应用时区分性不够明显。对此，提出VQ-MAP与LS—SVM融合的方法，使用通用背景模型（UBM）经过VQ—MAP过程得到说话人自适应参数集，把此参数集作为最小二乘支持向量机的训练样本应用于说话人识别系统中。用Matlab进行仿真实验，结果表明，该识别系统SVM训练时间短，且具有较高的识别率。

英文摘要：

Feature vectors used as the training samples of the traditional least square support vector machines does not give e- nough information to discriminate the voice in speaker recognition system. To solve this problem,this paper proposes the method based on VQ-MAP and LS-SVM. Adaptive parameter sets are got through VQ-MAP procedure using universal background model and are used as the training samples of LS-SVM in speaker recognition system. According to the results of simulation using Mat- lab, speaker recognition system based on VQ-MAP and LS-SVM uses less the training time of SVMs and it also has high recognition rate.