Study the influence of gender and age in recognition of emotions from algerian dialect speech

Abstract

Speech emotions Recognition is a very interesting area of research. In this work, the influence of gender and age on the speech emotions recognition in Algerian Dialect is studied. A n d on the other hand, the influence of speech emotion types on the classification of gender and age is also studied. An Algerian Dialect Emotional Datahase (A D E D) is used in this work. A D E D database is exploited for extracting the features that used in the systems of recognition and classification. These features are the statistic values of pitch and intensity, unvoiced frames, jitter, shimmer, H N R and M F C C s parameters. Analyzes based on gender and age are made to detect the influence of the four emotions on the parameters extracted. A parallel classifier composed of three classifiers, Support Vector Machines (S V M), K-Nearest Neighbor (K N N), and Linear Discriminant Analysis (L D A) is used in the recognition and classification systems. The results obtained show us that the performance of emotions recognition systems is influenced by gender and age i.e. the distinction between each gender class and each age interval in the recognition systems improves the performance compared to the systems without distinction. It was showed in the results also that the classifications of gender classes and age intervals were strongly influenced by the type of emotion.