The Application of Discriminative Training Techniques in LID System Fusion

摘要：

This paper reports an approach to language identification (LID) system fusion using discriminative training. Maximum mutual information (MMI) training for Gaussian mixture model is introduced to the standard LDA-GMM fusion framework. Experimental results show that the proposed fusion scheme outperforms the maximum likelihood (ML) trained backend of LID system. The impact of number of Gaussian mixtures on fusion performance is also discussed.

作者: Tao Hou Weiqiang Zhang Jia Liu

作者单位: Department of Electronic Engineering, Tsinghua University, Beijing 100084, China

会议类型: 国际会议

会议名称: 2008 International Conference on Audio，Language and Image Processing(2008国际声音、语言、图像过程大会)

会议地点: 镇江

会议语种:英文

页码: 1457-1460

在线出版日期: 2008-07-07（万方平台首次上网日期，不代表论文的发表时间）