Raga Classification and Recommendation for Healing Diseases Using Deep Learning
Gopa Pulastya, C. Lakshmi Sindhu, Rishi Anirudh K, B. H. Nanjunda Reddy, Suja Palaniswamy · 2025
Indian classical music, and specifically ragas, have been used for centuries with respect to healing. This research presents an artificial intelligence-based method for the automatic classification of Indian classical ragas along with mapping each of the classified raga to potential health benefits, with the aid of state-of-the-art deep learning methods. The framework uses Mel-Frequency Cepstral Coefficients (MFCCs) for feature extraction and incorporates data augmentation methods to improve the robustness of the model. A few machine learning models, namely Random Forest, K-Nearest Neighbors (KNN), and Convolutional Neural Networks (CNN), are used, with the outputs of all combined in a Voting Classifier. The top-performing model, a custom CNN, performs with an accuracy of 98.64%. Even though this work investigates the overlap of AI and music therapy, the claims of therapeutic uses rest on conventional believes and not scientific proofs. Major constraints include the lack of deployment in real-world situations and clinical proof. Future research will involve testing the system with real-users, scientifically proving therapeutic claims using clinical studies and investigating other deep learning architectures for better performance.