2026-05-12T23:53:54.611Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:1:4657 2017-01-09T16:20:24Z 1

Performance Analysis of Vocal Emotion Recognition Using Selective Speech Features 2016 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/1/4657 A. NJeguš М. Марјановић-Јаковљевић G. Anbarjafari Performance optimization in vocal emotion recognition is a challenging task in the field of vocal based human-computer interaction. In this paper, we apply three different classifiers for automatic vocal recognition, namely, Naïve Bayes, Support Vector Machine and Multilayer Perceptron. In order to select the most significant features, different filter feature selection strategies are applied. Within this approach, from 84 extracted state-of-the-art voice quality features, 23 features are selected, which are resulting in better recognition rate performance. In the experimental results Serbian language voice corpora is used. conferenceObject 1 5 Proceedings of 3rd International Conference on Electrical, Electronic and Computing Engineering - IcETRAN 2016