An Improved Method of Raga Content Assessment in Songs with Therapeutic Benefits

Abstract

A raga, in Indian classical music, be it Hindustani or Carnatic, is a melodic structure with fixed notes and a set of rules that characterize a particular mood conveyed by performance. Ragas have well acknowledged rich emotional content (the so called raga rasa) but a common man generally lacks the aesthetic sense to understand these ragas and hence cannot derive their therapeutic benefits completely. Accordingly when it comes to music medicine and music therapy, a nice idea would be to first render songs based on a particular raga one by one with gradually increasing raga content in them and then render this specific raga itself whereby determining the raga content in a song becomes an interesting and intriguing research problem. In an earlier work, we proposed a technique relying on correlation coefficient (based on pitch that characterize a musical note) of melodic shapes, as a measure of melodic similarity, to address the issue. Considering the limitations of correlation coefficient, in the present work, we propose an improved method by incorporating more features into our analysis. These additional features include note duration, inter onset interval and pitch movements between the notes. The experimental results are encouraging.