Abstract: Emotion recognition has been studied in recent past with great interest, out of various modalities from which emotions can be extracted, speech is the most natural and fastest of all the modalities. The major challenges for making a Speech emotion recognition system are finding and preparing database, selecting the most suitable features and designing appropriate classification scheme. The common approach is to extract a very large set of features over a generally long analysis time window and perform machine learning methods for classification. Speech emotion recognition increases the naturalness in Human Computer Interaction and can be used in wide variety of application in our day to day life. This paper surveys the three main building blocks of speech emotion recognition system, first part is survey of existing databases, the second part surveys most widely used features and the third part discuss various classification techniques.
Keywords: Emotion recognition, Speech emotion recognition, Statistical classifiers, Emotional speech databases.