Automatic carnatic raga identification using octave mapping and note quantization

This paper proposes a method to identify Carnatic Melakartha (Parent) ragas from audio recordings of Carnatic Classical music. Melakartha ragas are a set of seventy two ragas from which, all other ragas in the Carnatic music system are derived. The proposed method was classified into an Acoustic Model and a Music Language Model. In the Acoustic Model, a modified autocorrelation method was used, which detects the pitch frequency of the musical notes played, the pitch frequency being the acoustic feature. In the Music Language Model, the detected pitch frequencies were mapped from different musical octaves to the middle octave (octave 4 or C4) and then quantized to the standard frequencies to account for the characteristic variations of swaras in Carnatic music. From these frequencies, the unique frequencies with the highest occurrences were calculated and the raga was identified. Automatic raga identification can serve as a tutoring aid for learning musicians; can be used as an unbiased, performance assessment tool for vocalists/instrumentalists and for Music Information Retrieval in Carnatic music. The algorithm was tested on recordings of human vocals, the veena and the piano. The accuracy of raga identification for these recordings was 80.56%, 81.25% and 89.81% respectively.