Emotions play an important role in human perception and decision making. For a long time research on emotion intelligence has been done in the fields of psychology and cognitive science. Speech as the most important media of human communication contains a lot of emotional information, and how to automatically recognize speakers' emotional state has attracted many researchers' attention from different fields. In this paper, we presented a comparison of three different classification algorithms for detecting emotion from Mandarin speech. The results show that the proposed HMMs outperforms the other two classification techniques: about 3-6% improvement for K-NN and LDA.