EmoMeter: Measuring mixed emotions using weighted combinational model

Abstract

Emotion Recognition is an important area of affective computing and has potential applications. This paper proposes a combinational model to compute the percentage of different emotions jointly present in a given speech input. This model is a weighted combination of the classifier models like Neural Network, k-Nearest Neighbors, Gaussian Mixture Model, Naïve Bayesian Classifier and Support Vector Machines is proposed. The results of classification from the individual models are reported and compared with the proposed combinational model. It shows that the best performance is achieved using the proposed combination than the individual models.