Announcement for Downloading full text filePlease respect the Copyright Act.
All digital full text dissertation and theses from this website are authorized the copyright owners. These copyrighted full-text dissertation and theses can be only used for academic, research and non-commercial purposes. Users of this website can search, read, and print for personal usage. In respect of the Copyright Act of the Republic of China, please do not reproduce, distribute, change, or edit the content of these dissertations and theses without any permission. Please do not create any work based upon a pre-existing work by reproduction, Adaptation, Distribution or other means.
URN etd-0908115-174449 Statistics This thesis had been viewed 1357 times. Download 350 times. Author Yi-Ting Hung Author's Email Address No Public. Department Computer Science and Enginerring Year 2014 Semester 2 Degree Master Type of Document Master's Thesis Language zh-TW.Big5 Chinese Page Count 52 Title A Study on the Recognition of Specified Emotion from Continuous Speech Signals Keyword Continuous Speech Speech Recognition Emotion Recognition Emotion Recognition Speech Recognition Continuous Speech Abstract Speech emotion recognition is a process to recognize the emotion of the speaker from the uttered speech signal. To be more practical, it is necessary to use the natural dialogues as the training and testing corpus for continuous speech emotion recognition. Specific emotions recognitions from continuous speech can assist cell center systems, life-line and other telephone services. In order to reduce the computation time on practical applications and improve the efficiency, we reduce the number of features in the recognition process while maintain an acceptable recognition rate. In this research, we use two different cropora, Berlin Database of Emotional Speech(Emo-DB) and Mandarin Chinese Emotional Corpus 2010 (MCEC2010), and select the two most representative features form the thirteen features to establish specific emotion recognition models. In the experiment, we segment the continuous speech, then use the Deep Neural Networks(DNN) as classifier. The average recognition rate is 83.78%. In the experiment with conversational utterance in the MCEC2010 corpus database, the average recognition rate is 92.6%. Advisor Committee Tsang-Long Pao - advisor
Ching-Kuen Lee - co-chair
Yu Taso - co-chair
Files Date of Defense 2015-07-17 Date of Submission 2015-09-08