Emotion recognition from speech has developed as a recent research area in Human Computer Interaction (HCI). This paper is to use a BPNN classifier to classify four different emotions from 10 human user's speech database. This paper is to identify the users for which the data base has been trained and then identifying their emotions. In this, the authors present text independent speaker recognition. Speech emotion detection refers to discovering the speech category-based on the training and testing to the database provided.