AI Meets Music Machine Leaning for Indian Classical Raaq Classificatiðn

---

Abstract

Hindustani Classical Music is built upon intricate melodic structures known as Raags, each defined by unique tonal, temporal, and stylistic elements. Their recognition requires nuanced understanding and years of training, making automated classification a challenging task. This paper presents a machine learning framework for identifying Raag Yaman and Raag Malkauns, both standardized to A# pitch for consistency. Primary audio samples were recorded in a controlled studio environment, pre-processed to extract frequency-based statistical features (mean, median, and standard deviation), and used to train a Random Forest classifier. The proposed model achieved 91% accuracy in differentiating between the two Raags, demonstrating that simple statistical features, when paired with an ensemble learning approach, can yield high performance. The study contributes a reproducible, data-driven method for computational Raag analysis and highlights its potential in music education, digital archiving, and AI-assisted performance evaluation.

---

Keywords

Hindustani Classical Music, Machine Learning, Audio Processing, Random Forest Classifier