Carnatic raga identification using Arohana-avarohana pattern

Computer Science, Music Information Retrieval

Raga is a melodic framework for Indian classical music; it is a complex system consisting of notes, scales, and pitches. Raga mining is the technique of identifying the Raga of a song. Ragas can be distinguished from one another by their sequence of notes, melody, and its rendition. Identifying the Raga of a song can be challenging for beginners and non-musicians. Identifying the Raga can help learners classify songs to their corresponding Raga. This paper proposes the method of identifying the Raga by its Arohana-avarohana pattern. Arohana-avarohana pattern is the arrangement of notes occurring in the song ascending and descending order. The input is taken in the form of .wav file. The system converts the song into a file buffer. File buffer contains the information of the audio. This file buffer can be used to divide the song into segments, of 5 seconds and extract the notes in the segments. This is done using the Aubio library. The extracted notes are matched against the Arohana-avarohana pattern of the Ragas. The estimation of Raga is done based on the scoring method. For each note found, and matched against a Raga, the score for that Raga is incremented. The final Raga is found to be the Raga having the maximum score. This system was tested for various audio recordings of classical music and movie songs for the corresponding Raga. Two Ragas- Hindolam and Anandabhairavi are identified with an accuracy of 87%.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.