现代电子技术2016，Vol.39Issue(14)：1-3,8,4.DOI:10.16652/j.issn.1004-373x.2016.14.001
基于i-vector说话人识别算法中训练时长研究
Research on training duration of speaker recognition algorithm based on i-vector
摘要
Abstract
In order to further enhance the system performance of speaker recognition model based on i⁃vector,the effect of the training time,the sex ratio and Gaussian mixture in the speaker recognition system based on i⁃vector on the performance of system identification is discussed in this paper. In this article,a set of experiments is set in allusion to the training time,the sex ratio and Gaussian mixture,and is verified in combination with the most popular voice recognition tool Kaldi. The best pa⁃rameters of the speaker recognition algorithm based on i⁃vector were obtained,which provided a data basis for later research on the speaker recognition algorithm based on i⁃vector.关键词
说话人识别/i-vector/Kaldi/训练时长Key words
speaker recognition/i-vector/Kaldi/training duration分类
电子信息工程引用本文复制引用
马平,黄浩,程露红,杨萌萌..基于i-vector说话人识别算法中训练时长研究[J].现代电子技术,2016,39(14):1-3,8,4.基金项目
国家自然科学基金 （）
