出版社：Vilnius University, University of Latvia, Latvia University of Agriculture, Institute of Mathematics and Informatics of University of Latvia
摘要：Various feature selection and classification schemes were proposed to improve efficiency of speech emotion classification and recognition. In this paper we propose multi-level organization of classification process and features. The main idea is to perform classification of speech emotions in step-by-step manner using different feature subsets for every step. We applied the maximal efficiency feature selection criterion for composition of feature subsets in different classification levels. The proposed multi-level organization of classification and features was tested experimentally in two emotions, three emotions, and four emotions recognition tasks and was compared with conventional feature combination techniques. Using the maximal efficiency feature selection criterion 2nd and 16th order multi-level feature sets were composed for three and four emotions recognition tasks respectively. Experimental results show the superiority of proposed multi-level classification scheme by 6,3–25,6 % against straightforward classification and conventional feature combination schemes.
