| 基于DTW的编码域说话人识别研究 |
| |
| 引用本文： | 李榕健,于洪涛,李邵梅.基于DTW的编码域说话人识别研究[J].电子技术应用,2010,36(8). |
| |
| 作者姓名： | 李榕健 于洪涛 李邵梅 |
| |
| 作者单位： | 国家数字交换系统工程技术研究中心; |
| |
| 基金项目： | 国家863计划重点项目 |
| |
| 摘 要： | 相对解码重建后的语音进行说话人识别,从VoIP的语音流中直接提取语音特征参数进行说话人识别方法具有便于实现的优点,针对G.729编码域数据,研究基于DTW算法的快速说话人识别方法。实验结果表明,在相关的说话人识别中,DTW算法相比GMM在识别正确率和效率上有了很大提高。
|
| 关 键 词： | 说话人识别 G.729 编码流 语音特征参数 DTW算法 |
Research on speaker recognition from coding domain based on DTW algorithm |
| |
| LI Rong Jian,YU Hong Tao,LI Shao Mei.Research on speaker recognition from coding domain based on DTW algorithm[J].Application of Electronic Technique,2010,36(8). |
| |
| Authors: | LI Rong Jian YU Hong Tao LI Shao Mei |
| |
| Affiliation: | LI Rong Jian,YU Hong Tao,LI Shao Mei(National Digital Switching System and Researching Center,Zhengzhou 450002,China) |
| |
| Abstract: | The direct extraction of speech feature parameters from VoIP voice stream for speaker recognition in relation to reconstructing the decoded speech for speaker recognition has the advantages of easy implementation.This paper focuses on data in G.729 coding domain and proposes a method of quick speaker recognition based on DTW algorithm.Experimental results show that,compared with GMM,DTW has advantage of high rcognition accuracy and process speed in the text-dependent speaker recognition. |
| |
| Keywords: | speaker recognition G 729 coding stream voice feature parameter DTW algorithm |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
