Modeling of Acoustic emotion recognition using Artificial Intelligence and Machine Learning

Abstract

The emotion is a kind of language that can be understand by speech. If a machine can understand the emotions by its intelligence, then it refers as artificial intelligent. Therefore, in this paper we proposed an artificial intelligence technique for recognizing acoustic emotions. In this paper, the modeling of acoustic emotion is done by the fusion of classifiers such as MLP, SVM, KNN, Random forest and voting classifier. The voting can be ‘Soft’ and ‘Hard’. In hard voting the output is proportional to the highly voted or favorable class where as in soft voting the output is proportional to average voting. The best combination that we found with the fusion of MLP, SVM, Random Forest classifiers. The voting in this case was soft voting with the accuracy of 88.09%. It is higher than any of the single classifier. The proposed model is executed on the standard datasets i.e. Ravdess dataset.