| 基于参考说话人模型和双层结构的说话人辨认 |
| |
| 引用本文： | 王刚,邬晓钧,郑方,王琳琳,张陈昊.基于参考说话人模型和双层结构的说话人辨认[J].清华大学学报(自然科学版),2011(9):1261-1266. |
| |
| 作者姓名： | 王刚 邬晓钧 郑方 王琳琳 张陈昊 |
| |
| 作者单位： | 清华信息科学技术国家实验室技术创新与开发部语音和语言技术中心;清华大学计算机科学与技术系; |
| |
| 摘 要： | 为了提高基于Gauss混合模型通用背景模型(GMM-UBM)的说话人辨认系统的运算效率,提出一种基于参考说话人模型的双层结构用于目标说话人剪枝,采用矢量量化方法从目标说话人模型集合中训练参考说话人模型,利用语音与参考说人模型的偏差来描述说话人的发音特性,将辨认语音偏差向量和目标说话人偏差向量的相似性作为距离度量来进行目标说话人剪枝。实验结果表明:在基于GMM-UBM的说话人辨认系统中,对包含5 200个目标说话人和1 000个集外说话人的测试集进行开集辨认的条件下,在提高辨认的运算效率12.5倍的同时识别率仅下降0.3%。
|
| 关 键 词： | 双层结构 快速说话人辨认 参考说话人模型 |
Speaker identification using a reference speaker model based a two-layer structure |
| |
| WANG Gang,WU Xiaojun,ZHENG Thomas Fang,WANG Linlin,ZHANG Chenhao.Speaker identification using a reference speaker model based a two-layer structure[J].Journal of Tsinghua University(Science and Technology),2011(9):1261-1266. |
| |
| Authors: | WANG Gang WU Xiaojun ZHENG Thomas Fang WANG Linlin ZHANG Chenhao |
| |
| Institution: | WANG Gang,WU Xiaojun,ZHENG Thomas Fang,WANG Linlin,ZHANG Chenhao(1.Center for Speech and Language Technologies,Division of Technical Innovation and Development,Tsinghua National Laboratory for Information Science and Technology,Beijing 100084,China,2.Department of Computer Science and Technology,Tsinghua University,China) |
| |
| Abstract: | The Gaussian mixture model-universal background model(GMM-UBM) based speaker identification system's computation efficiency is improved by a fast algorithm using a reference speaker model based two-layer structure.Vector quantization was used to train the reference speaker models using target speaker models.The deviations between one speaker and the reference speaker models were used to model the speaker's acoustic characteristics.The correlation between the deviation vectors was used to evaluate the simila... |
| |
| Keywords: | two-layer structure fast speaker identification reference speaker model |
| 本文献已被 CNKI 等数据库收录！ |
|
