| 基于GMM的说话人辨认系统及其改进 |
| |
| 引用本文： | 谢霞,李宏,郑俊. 基于GMM的说话人辨认系统及其改进[J]. 电脑与信息技术, 2006, 14(2): 48-51 |
| |
| 作者姓名： | 谢霞 李宏 郑俊 |
| |
| 作者单位： | 中南大学信息科学与工程学院,湖南,长沙,410083;中南大学信息科学与工程学院,湖南,长沙,410083;中南大学信息科学与工程学院,湖南,长沙,410083 |
| |
| 摘 要： | 建立声学模型是说话人识别技术的重要环节.文章介绍了一种改进的GMM算法,将基于样本和核的相似性度量的动态聚类算法与传统高斯混合模型结合起来进行建模,识别辨认时,对语音帧得分进行加权处理.实验表明：改进后的与文本无关的说话人辨认系统无论是在建模时间还是识别效率上都要高于传统的基于GMM的说话人辨认系统.
|
| 关 键 词： | 混合高斯模型 帧似然概率 聚类算法 |
| 文章编号： | 1005-1228(2006)02-0048-04 |
| 收稿时间： | 2006-01-07 |
| 修稿时间： | 2006-01-07 |
GMM-based speak identification system and its improvement |
| |
| XIE Xia,LI Hong,ZHENG Jun. GMM-based speak identification system and its improvement[J]. Computer and Information Technology, 2006, 14(2): 48-51 |
| |
| Authors: | XIE Xia LI Hong ZHENG Jun |
| |
| Affiliation: | School of Information Science and Engineering, Central South University, Changsha,Hunan 410083, China |
| |
| Abstract: | Characteristic modeling is an important link in technology of speaker identification.This article introduces an improved algorithm of GMM which combines classical GMM with clustering algorithm based on similarity measure between stylebook and nuclear for modeling and weighting speech frames' score while recognizing.From the results,we can conclude that the improved uncertain text-speaker-recognition system has higher performance than classic system both on modeling velocity and recognition rate. |
| |
| Keywords: | Gaussian Mixture Model frame likelihood clustering algorithm |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
