Computer Engineering and Applications ›› 2007, Vol. 43 ›› Issue (10): 41-43.

• 学术探讨 • Previous Articles     Next Articles

Automatic Language Identification using SVM-UBM

  

  • Received:2006-08-14 Revised:1900-01-01 Online:2007-04-01 Published:2007-04-01

基于SVM-UBM的语言辨识系统

张文林 李弼程 屈丹   

  1. 郑州 信息工程大学信息工程学院信息科学系 解放军信息工程大学
  • 通讯作者: 张文林

Abstract: As powerful theoretical and computational tools, support vector machines (SVMs) have been widely used in pattern classification of many areas. In this paper, we present a general framework for language identification using SVMs, introduce the use of Louradour sequence kernel into language identification system, and develop a universal background Gaussian Mixture Model to improve it’s performance. Experiment results demonstrate that the SVM-UBM system not only yields performance superior to those of a GMM classifier but also outperforms the system using Generalized Linear Discriminant Sequence (GLDS) Kernel.

Key words: language identification, support vector machine, sequence kernel, gaussian mixture model, universal background model

摘要: 支持向量机作为强大的理论工具和计算工具,已成功地应用在模式识别的众多领域中。本文研究了将支持向量机模型(SVM)应用于语言辨识的理论框架,提出了将Louradour序列核应用于语言辨识,并利用高斯混合模型(GMM)构造全局背景模型(UBM)对其进行了改进,从而导出了基于SVM-UBM的语言辨识系统。相关实验结果表明,该系统的识别率高于经典的高斯混合模型(GMM)和基于广义线性区分性核(GLDS)的支持向量机模型。

关键词: 语言辨识, 支持向量机, 序列核, 高斯混合模型, 全局背景模型