The Investigation the Effects of the Performance of an Independent Emotion Recognition of Model Used in the Dimensioning of Emotions

Abstract

Emotion recognition aims at determining the state of emotion which is included in the speech or mimics of a person. Emotion recognition from speech is an area related to signal processing and psychology. Acoustic parameters obtained from speech signals through acoustic analysis, which is one of objective evaluation methods, is intensively used in emotion recognition studies. In this paper success in emotion recognition is examined in categorical aspects and the impact of dimensional model on independent emotion recognition success is investigated. Acoustic parameters were subjected to classification with Support Vector Machine in order to determine emotion recognition success. According to the obtained findings, emotion recognition success in categorical structure, dimensional structure and categorical-dimensional structure were 69.5 percent, 73.3 percent and 87.1 percent, respectively. Even if dimensional structure is higher in arousal than in valence, when emotion recognition success is examined on each emotion dimension, valence provided higher success.