版权所有:内蒙古大学图书馆 技术提供:维普资讯• 智图
内蒙古自治区呼和浩特市赛罕区大学西街235号 邮编: 010021
作者机构:Univ Politecn Valencia DSIC MLLP E-46022 Valencia Spain
出 版 物:《COMPUTER SPEECH AND LANGUAGE》 (计算机语音与语言)
年 卷 期:2016年第37卷
页 面:11-23页
核心收录:
学科分类:08[工学] 0812[工学-计算机科学与技术(可授工学、理学学位)]
基 金:European Union Spanish MINECO [TIN2009-14511, TIN2012-31723] FPI Scholarship [BES-2010-033005]
主 题:Confidence measures Speech recognition Speaker adaptation Log-linear models Online video lectures
摘 要:Automatic speech recognition applications can benefit from a confidence measure (CM) to predict the reliability of the output. Previous works showed that a word-dependent native Bayes (NB) classifier outperforms the conventional word posterior probability as a CM. However, a discriminative formulation usually renders improved performance due to the available training techniques. Taking this into account, we propose a logistic regression (LR) classifier defined with simple input functions to approximate to the NB behaviour. Additionally, as a main contribution, we propose to adapt the CM to the speaker in cases in which it is possible to identify the speakers, such as online lecture repositories. The experiments have shown that speaker-adapted models outperform their non-adapted counterparts on two difficult tasks from English (***) and Spanish (poliMedia) educational lectures. They have also shown that the NB model is clearly superseded by the proposed LR classifier. (C) 2015 Elsevier Ltd. All rights reserved.