Clustering the Raagas of Sankari Sangeet —A Computational Approach

Abstract

Raaga is the heart of Indian Classical Music. A raaga is an arrangement of 12 notes in the octave. Like Indian classical music, Sankari Sangeets which are composed 500 years ago are also based on raaga. The musicologist of Sankari Sangeet found nearly 26 different raagas in Sankaree Sangeet. Each of these raagas has some specific characteristics such as Aruhana (Ascending order of notes), Avaruhana (Descending order of notes), Pakad, Time of Singing, Vadi swara (Dominant/Important Note), Samvadi swara (Second Dominant Note), etc. This piece of work considers all of the 26 raagas of Sankari Sangeet to statistically cluster on the basis of 18 features extracted from each raaga. For clustering of the raagas, k-prototypes clustering technique is used which gives four clusters of sizes 9, 4, 6, and 7. Also, a comparison is carried out between the clusters obtained from k-prototypes clustering and existing theoretical Thaat-based classification.