结合上下文多维度声学特征组的汉语重音检测  

The Detection of Mandarin Stress Combined with Context Multidimensional Acoustic Features of the Group

作　　者：赵云雪[1] 张珑[1] 郑世杰[1] 

机构地区：[1]哈尔滨师范大学

出　　处：《哈尔滨师范大学自然科学学报》2014年第3期93-98,共6页Natural Science Journal of Harbin Normal University

基　　金：黑龙江省应用技术研究与开发计划项目(GZ13A003)

摘　　要：基于ASCCD朗读语篇语料库,结合上下文多维度声学特征组,并采用NaiveBayes分类器对汉语重音进行检测.NaiveBayes分类器对当前音节的相关声学特征及上下文的相关特征进行建模,这种分类方法充分利用了当前音节及上下文的特性.实验结果证明该声学特征组在ASCCD朗读语篇语料库上能够得到80.7%的汉语重音检测正确率.In this paper, the stress are conducted by using acoustic features of the group and NaiveBayes classifier based on ASCCD reading discourse corpus. NaiveBayes classifier is utilized to model the acoustic , which adequately utilizes the property of the current syllable and context. The experimental results indicate that the acoustic features of the group achieves 80.7% stress detection accuracy on ASCCD.

关 键 词：重音 NaiveBayes分类器 声学特征 上下文特征 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]