融合上下文短时谱特征的汉语重音检测研究  

Chinese Accent Detection Research Integration Context Short-time Spectrum Features

作　　者：赵云雪[1] 张珑[1,2] 郑世杰[1] 

机构地区：[1]哈尔滨师范大学计算机科学与信息工程学院,哈尔滨150025 [2]哈尔滨工业大学计算机科学与技术学院,哈尔滨150001

出　　处：《智能计算机与应用》2014年第4期36-38,共3页Intelligent Computer and Applications

基　　金：黑龙江省自然科学基金(F201321);黑龙江省应用技术研究与开发计划项目(GZ13A003)

摘　　要：重音是语言交流中不可或缺的部分,在语言交流中扮演着非常重要的角色。本文基于ASCCD朗读语篇语料库,使用MFCC算法提取每个语音段的融合上下文子段拼接短时谱信息,构建基于MFCC算法的上下文短时谱特征集;并选用NaiveBayes分类器对这类特征集进行建模,而且将具有最大后验概率的类作为该对象所属的类,这种分类方法充分利用了当前语音段的相关语音特性;融合上下文的MFCC短时谱特征组在ASCCD上能够得到83.6%的汉语重音检测正确率。实验结果证明,融合上下文子段拼接特征规整方法可以用于汉语重音检测研究中。Accent is a critically important component of spoken communication, and plays a very important role in spoken communication. This paper selects from ASCCD corpus and conducts accent by using MFCC algorithm to extract each voice segment of short - time spectrum based on context sub - segment splicing information. After that, the paper builds integration context short -time spectrum feature sets based on MFCC algorithm, and chooses NaiveBayes classifier to model the two feature sets. NaiveBayes is to choose the classes with maximum a posteriori probability as the object＇s class. This classification method makes full use of the related phonetic features of speech segment. Integration context short - time spectrum of MFCC feature set respectively achieves 83.6% accent detection accuracy on ASCCD. The experimental results indicate that integration context sub - segment splicing feature structured method of MFCC can be used in Chinese accent detection study.

关 键 词：重音 上下文 短时谱特征 重音检测 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]