Deep learning model for speech emotion classification based on GCI and GOI detection

The temporal regions of considerable stimulation to the vocal tract during the production of spoken speech are referred to as glottal closure instants (GCIs) and glottal opening instants (GOIs). These instants from speech signals are important for various speech and biomedical applications. The performance of the GCI and GOI detection algorithm is not studied for emotion speech using deep learning technique. Hence, the motivation behind this work is to use the current edge technology. To the best of our knowledge ours is the first deep learning CNN model to classify different emotions of speech based on GCI and GOI. In this work, an attempt is made to study the performance for different emotions using GCI and GOI detection algorithm and make the comparison with the state of art methods. The performance of the developed model is evaluated using traditional evaluation metrics. The results show that GCIs and GOIs are helpful in classification of speech emotion and that performance of the proposed model decreases for anger and sad emotions.