SRiyam Net: Exploring Emotion Recognition in Hindustani Classical Music

The recognition of emotions in Hindustani classical music has gained significant interest due to its potential applications in music therapy, performance enhancement, and human machine interaction. In this research, we propose SRiyam Net, an innovative emotion classification model specifically tailored to the unique characteristics of Hindustani classical music. We conducted a comprehensive evaluation of SRiyam Net, comparing its performance with a conventional state-of-the-art CNNBiL-STM model across various metrics. SRiyam Net demonstrated superior results with higher precision, recall, and F1 scores for both the "Happy" and "Sad" classes, achieving an overall accuracy of 0.76. In contrast, the state-of-the-art model showed lower performance, with lower precision, recall, and F1 scores, resulting in an accuracy of 0.52.