Acoustic Feature Identification to Recognize Rag Present in Borgit

Abstract

In the world of Indian classical music, raga recognition is a crucial undertaking. Due to its particular sound qualities, the traditional wind instrument known as the borgit presents special difficulties for automatic raga recognition. In this research, we investigate the use of auditory feature identification methods to create a reliable raga recognition system for Borgit performances. Each of the Borgits, the devotional song of Assam is enriched with rag and each rag has unique melodious tune. This paper has carried out few experiments on the audio samples of rags and a few Borgits sung with those rugs. In this manuscript three mostly used rags and a few Borgits with these rags are considered for the experiment. Acoustic features considred here are FFT (Fast Fourier Transform), ZCR (Zero Crossing Rates), Mean and Standard deviation of pitch contour and RMS(Root Mean Square). After evaluation and analysis it is seen that FFT and ZCR are two noteworthy acoustic features that helps to identify the rag present in Borgits. At last K-means clustering was applied on the FFT and ZCR values of the Borgits and were able to find correct grouping according to rags present there. This research validates FFT and ZCR as most precise acoustic parameters for rag identification in Borgit. Here researchers had observed roles of Standard deviation of pitch contour and RMS values of the audio samples in rag identification.