| 基于FVQMM的说话人识别 |
| |
| 引用本文： | 成新民,张迎,蒋云良.基于FVQMM的说话人识别[J].辽宁工程技术大学学报(自然科学版),2007,26(5):719-722. |
| |
| 作者姓名： | 成新民 张迎 蒋云良 |
| |
| 作者单位： | 1. 湖州师范学院,信息工程学院,浙江,湖州,313000
2. 浙江大学,计算机科学与技术学院,浙江,杭州,310027 |
| |
| 基金项目： | 国家自然科学基金资助项目(60573056) |
| |
| 摘 要： | 为了克服传统VQ与GMM说话人识别的缺点,提出了一种新的FVQMM说话人识别方法。该方法综合了VQ、GMM和模糊集理论的优点。通过用模糊VQ误差尺度取代传统GMM的输出概率函数,减少了建模时对训练数据量的要求,提高了识别速度。实验结果表明该方法是有效的。
|
| 关 键 词： | 说话人识别 矢量量化 混合高斯模型 |
| 文章编号： | 1008-0562（2007）05-0719-04 |
| 修稿时间： | 2006-03-24 |
Speaker recognition based on FVQMM |
| |
| CHENG Kin-min,ZHANG Ying,JIANG Yun-liang.Speaker recognition based on FVQMM[J].Journal of Liaoning Technical University (Natural Science Edition),2007,26(5):719-722. |
| |
| Authors: | CHENG Kin-min ZHANG Ying JIANG Yun-liang |
| |
| Institution: | 1.School of Information Engineering, Huzhou Normal College Huzhou 313000, China; 2.College of Computer Science, Zhejiang University, Hangzhou 310027 ,China |
| |
| Abstract: | In order to overcome the defects of the speaker reorganization of traditional VQ and GMM,a new method of speaker recognition is presented.When VQ error scale is adopted instead of the probability output by traditional GMM,data capacity request is reduced during modeling and reaches a high rate of recognition.The result also shows that the method is effective for speaker recognition. |
| |
| Keywords: | speaker recognition VQ gaussian mixture model |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
