An Approach of Similarity Measure in Hindustani Music

Music is a performing art. From the ancient time musical evaluation vis-a-vis perfection is subjective. Usual methods of assessment in vogue are subjective and depend largely on the personality and the background as well as the mood and temperament of the assessor. Therefore, a quantitative method for the assessment of the closeness of the performance of a student with respect to his/her teacher/idol in Hindustani classical music is a necessity. In this paper, we study the methods to evaluate the features which might estimate the closeness of two music signals. Here we compare some features extracted from analyzing the audio signals of both the teacher and the student and discuss how these features can be used as a measure of closeness. These are beat, note duration, the time between two successive notes, frequency of each note, frequency shift between successive notes, attack time of each note and mel frequency.