欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为了提高基于高斯混合模型-通用背景模型（GMM-UBM）说话人识别系统的运算速度,提出了通用背景模型（UBM）降阶算法,该方法采用极大似然估计法训练一个高阶UBM,再采用UBM降阶算法得到低阶UBM.采用最短距离高斯分量替换空映射集合的方法解决了空映射集问题.通过实验方法分析了3种初始化低阶UBM方法的识别结果,发现不同的初始化方法对结果影响很小.在NIST2001 SRE数据库上的实验显示,该算法使基于GMM-UBM说话人识别系统的运算速度提高了8倍,而等错误率仅上升了4.59%,表明了UBM降阶算法在小幅降低系统识别率的情况下,可大幅度提高GMM-UBM系统的运行效率.
英文摘要：
A universal background model （UBM） reduction method was proposed to speed up the Gaussian mixture model-universal background model （GMM-UBM） based speaker recognition system. A high-order UBM was trained by expectation maximization （EM） algorithm and then clustered into a new UBM with lower order. The Gaussian component with the shortest distance was adopted to replace the empty set to solve the empty mapping set problem. Three methods of initialization low-order UBM were experimentally analyzed to find out that different initialization methods converged to similar recognition results. The experiments on NIST2001 SRE Corpora showed that the equal error rate （EER） of the system only increased 4.59%, while the computation speed increased by 8 times. The UBM reduction method can considerably improve the efficiency of the GMM-UBM system while maintaining the performance.
同期刊论文项目
同项目期刊论文
期刊信息
位置：