基于隐马尔可夫模型的物联网终端语音身份动态识别方法  

Dynamic recognition method for voice identity of Internet of things terminal based on hidden Markov model

作　　者：郭霏霏 GUO Feifei(Institute of Intelligent Manufacturing,Quanzhou Vocational and Technical University,Quanzhou 362000,Fujian,China)

机构地区：[1]泉州职业技术大学智能制造学院,福建泉州362000

出　　处：《上海电机学院学报》2021年第6期361-365,共5页Journal of Shanghai Dianji University

基　　金：2020年福建省本科高校教育教学改革研究重大资助项目(FBJG20200007)。

摘　　要：为提高多角度面部表情识别的精度,提出了一种基于隐马尔可夫模型的物联网终端语音身份动态识别方法。通过隐马尔可夫函数特征,构建物联网终端语音模型和身份模型,结合两种模型完成隐马尔可夫语音身份特征的建模。在此基础上,构建物联网终端语音的采集平台,完成物联网终端语音数据的采集,再对原始信号进行滤波预处理,使用加窗方法提取处理后语音信号的身份特征,并对身份特征信号数据进行分类识别,从而完成基于隐马尔可夫的物联网终端语音身份动态识别。实验结果表明,该方法识别精准度较高,识别用时较短,且稳定性较好。In order to improve the accuracy of multi-angle facial expression recognition,a dynamic speech identity recognition method for Internet of things terminals is proposed based on Hidden Markov model.The speech model and identity model of Internet of things terminal are constructed through the hidden Markov function features,and the modeling of the hidden Markov speech identity feature is completed by combining the two models.On this basis,the voice acquisition platform of the Internet of things terminal is constructed to complete the voice data acquisition of the Internet of things terminal.Then the original signal is filtered and preprocessed,the windowing method is used to extract the identity features of the processed voice signal,and the identity feature signal data is classified and recognized.Therefore,the dynamic voice identity recognition of the Internet of things terminal is completed based on hidden Markov.The experimental results show that this method has higher recognition accuracy,shorter recognition time and better stability.

关 键 词：隐马尔可夫 语音采集 身份识别 滤波 加窗 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]