A DCRNN-based ensemble classifier for speech emotion recognition in Odia language

Abstract The Odia language is an old Eastern Indo-Aryan language, spoken by 46.8 million people across India. We have designed ensemble classifier using Deep Convolutional Recurrent Neural Network for Speech Emotion Recognition (SER). This study presents a new approach SER tasks motivated recent research on speech emotion recognition. Initially, we extract utterance-level log Mel-spectrograms and their first second derivative (Static, Delta, Delta-delta), represented as 3-D Mel-spectrograms. utilize deep convolutional neural networks to the features from Then bi-directional-gated recurrent unit network applied express long-term temporal dependency out of all produce emotion. Finally, use classifiers Softmax Support Vector Machine improve final recognition rate. In this way, our proposed framework trained tested (Seven emotional states) RAVDESS (Eight dataset. experimental results reveal that performs better instead single classifier. accuracy levels reached are 85.31% 77.54%, outperforming some state-of-the-art frameworks datasets.