Automatic Persian Speech Emotion Recognition: A CNN-Based Approach Using PDREC and ShEMO Datasets

The need for advanced Speech Emotion Recognition (SER) systems has grown with the development of human-machine interaction technologies. This paper introduces a CNN model specifically designed to detect emotions in Persian speech, using two well-known datasets: the Persian Drama Radio Emotional Corpus (PDREC) and the Sharif Emotional Speech Database (ShEMO). The CNN model achieved high accuracy rates of 84% on the PDREC dataset and 90.21% on the ShEMO dataset, showing its reliability in different conditions. These results highlight the model's potential for real-world SER applications, improving the quality of human-computer interactions.