| 与文本无关说话人识别 |
| |
| 引用本文： | 赵玉晓顾秀秀张二华. 与文本无关说话人识别[J]. 计算机与数字工程, 2014, 0(2): 243-247,307 |
| |
| 作者姓名： | 赵玉晓顾秀秀张二华 |
| |
| 作者单位： | 南京理工大学计算机科学与技术学院,南京210094 |
| |
| 摘 要： | 由于传统的说话人识别中,常用的特征参数有线性预测系数（LPC）、Mel频率倒谱系数（MFCC）,采用单一特征参数并不能很好地反映说话人特性.针对这种情况,提出了引入Delta特征和特征组合的方法.实验结果表明,引入Delta特征和特征组合对识别效果有明显提高,实验中选用GMM作为说话人识别模型.
|
| 关 键 词： | 线性预测系数 梅尔倒谱系数 Delta特征 高斯混合模型 |
Test-Independent Speaker Recognition |
| |
| ZHAO YuxiaoCollege of Computer Science and Technology,Nanjing University of Science ＆ Technology,Nanjing,GU XiuxiuCollege of Computer Science and Technology,Nanjing University of Science ＆ Technology,Nanjing,ZHANG Erhua. Test-Independent Speaker Recognition[J]. Computer and Digital Engineering, 2014, 0(2): 243-247,307 |
| |
| Authors: | ZHAO YuxiaoCollege of Computer Science Technology Nanjing University of Science ＆ Technology Nanjing GU XiuxiuCollege of Computer Science Technology Nanjing University of Science ＆ Technology Nanjing ZHANG Erhua |
| |
| Affiliation: | (College of Computer Science and Technology, Nanjing University of Science ＆ Technology, Nanjing 210094) |
| |
| Abstract: | In traditional speaker recognition research,the commonly feature parameters include Linear Prediction Coefficients（LPC）、Mel-Frequency Cepstral Coefficients（MFCC） ect.One feature parameter cannot reflect the characteristics of the speaker well.Therefore,the Delta feature parameters and the feature parameters combination is proposed.Experimental results show that the recognition rate is improved obviously.GMM is used as the training model. |
| |
| Keywords: | linear prediction coefficients mel-frequency cepstral coefficients delta feacher gaussian mixture model |
| 本文献已被 CNKI 维普 等数据库收录！ |
