www.ijsrp.org Comparison of ML classifiers for Raga recognition

Abstract- Indian classical music is categorized into two major forms: Hindustani music and Carnatic music which are performed and practiced in North and South India. Raga is the fundamental concept in Indian music on which whole melody of a performance is based and comprises of group of swaras (musical notes) ranging from 5-8 notes. The present work focuses on Hindustani raga identification. We perform raga recognition on a set of ragas and collected dataset of live performances of these ragas, both vocal and instrumental, and performed recognition using machine learning classifiers. A comparative analysis is done between these classifiers based upon their parameters like average accuracy, precision, recall, kappa statistic and results are being formulated. The result shows that K-star algorithm performs better with raga identification with an accuracy of 93.38%