Two Step Recognition of Raags in Hindustani Classical Music Using Supervised Deep Learning

Abstract

In the Indian Classical system of music, we have various raags which differ from one another based on the notes being used in them, their frequency bandwidth, the subtle intricacies of the progression of notes, etc. In [1], we have worked upon the classification of a sample of music into a thaat which is an upper-level classification simply based on the frequencies present in it. In this paper, we take the work one step ahead to recognize the raag after successful classification of the thaat. In this way, it forms a two-step process where we first identify the thaat under which the music falls, followed by which we recognize the raag which corresponds to the music in consideration.