Abstract: In recent years, extensive concern about calculated speech emotion has been aroused at home and abroad. Especially, a lot of studies have been done in speech emotion feature extraction. The Intrinsic Mode Function（IMF） for the first four steps of the emotion speech is attained in this paper by using the method of Empirical Mode Decomposition（EMD） to process the emotion speech. And the instantaneous frequency and amplitude of the Intrinsic Mode Function（IMF） for the first four steps are got by Hilbert transformation respectively. The emotion characteristics vector is composed by extracting their statistical characteristics and combining the acoustic characteristics of emotion speech and characteristics vector is normalized. The four kinds of emotion speech, namely angry, happiness, sadness and calmness are recognized by using the Support Vector Machine（SVM）. Experimental results show that the recognition effect of this proposed method is much better.

Key words: Empirical Mode Decomposition（EMD）, feature extraction, Support Vector Machine（SVM）, emotion recognition

摘要： 语音情感计算引起了国内外广泛的关注，特别是在语音情感特征提取方面做了大量的研究。利用经验模态分解（EMD）方法对情感语音进行处理，得到情感语音的前4阶固有模态函数（IMF），并将前4阶IMF分别通过Hilbert变换得到其瞬时频率和瞬时振幅。提取它们的统计特征，再结合情感语音的声学特征共同组成情感特征向量，并对特征向量做归一化处理。利用支持向量机（SVM）对四种情感语音即生气、高兴、悲伤和平静进行识别。实验结果表明该方法的识别效果较好。

关键词: 经验模态分解（EMD）, 特征提取, 支持向量机（SVM）, 情感识别