期刊名称:International Journal of Engineering and Computer Science
印刷版ISSN:2319-7242
出版年度:2015
卷号:4
期号:6
页码:12523-12528
出版社:IJECS
摘要:In machine interaction with human being is yet challenging task that machine should be able to identify and react to humannon-verbal communication such as emotions which makes the human computer interaction become more natural. In present research areaautomatic emotion recognition using speech is an essential task which paid close attention. Speech signal is a rich source of informationand it is an attractive and efficient medium due to its numerous features of expressing approach & extracting emotions through speech ispossible. In this paper emotions is recognized through speech using spectral features such as Mel frequency cepstrum coefficient prosodicfeatures like pitch , energy and were utilized & study is carried out using K- Nearest Neighbor classifiers , Support Vector MachineClassifier and Gaussian mixture model classifier which is used for detection of six basic emotional states of speaker’s such as anger,happiness , sadness , fear , disgust and neutral using Berlin emotional speech database.
关键词:Classifier ;Emotion recognition; features generation ; spectral features; prosodic features.