Emotion Recognition Based on Speech Signal Using Hybrid Features with Decision tree

Abstract

Emotion recognition based on a speech signal is one of intensively studied research topics in the domains of human-computer interaction and affective computing. The main idea are a new Hybrid feature set was introduced in extract features , which use the basic concept in work is the Residual Signal of the prediction procedure, which is the difference between the original and its prediction ,by enter to the internal structure of the signal axis and calculate a features for some specific regions after that calculate moments for these regions . Machine learning techniques will be used to classify to achieve a high degree of accuracy.Publicly available speech datasets like the berlin dataset are tested using a decision tree classifier. The hybrid features were trained separately. The results indicated that features very encouraging, reaching 98.5%. In this article, the decision tree classifier test results with the same tested hybrid features that published in a previous article will be presented, also a comparison between some related works and the proposed technique in speech emotion recognition techniques.