Comparison of Classification Methods for Detecting Emotion from Mandarin Speech

摘要: It is said that technology comes out from humanity. What humanity? The very definition of humanity emotion. Emotion the basis for all human expression and underlying theme behind everything done, said, thought or imagined. Making computers being able to perceive respond emotion, human-computer interaction will be more natural. Several classifiers are adopted automatically assigning an emotion category, such as anger, happiness sadness, a speech utterance. These were designed independently tested on various emotional corpora, making it difficult compare evaluate their performance. In this paper, we first compared several popular classification methods evaluated performance by applying them Mandarin corpus consisting five basic emotions, including happiness, boredom, sadness neutral. extracted feature streams contain MFCC, LPCC, LPC. experimental results show proposed WD-MKNN classifier achieves accuracy 81.4% 5-class recognition outperforms other techniques, KNN, MKNN, DW-KNN, LDA, QDA, GMM, HMM, SVM, BPNN. Then, verify advantage method, these another expressive two emotions. still others.