Classification of Ragas of Indian Classical Music Using RNN-LSTM

In general, a raga may be thought of as a musical entity that contains note intonation, relative length, and order. It is a combination of technical and ideational elements present in music. The idea offers versatility, allowing the artist to use straightforward expression or add decoration while still conveying the same key ideas and invoking various moods. Some guidelines are seen as mandatory while others are optional. This paper gives a system and a tool for classifying distinct ragas in a particular music composition from the perspective of musicians, artists, and writers of Indian Classical Music Raga Compositions. This system is beneficial not just to the artists, but also to new Carnatic or Hindustani music students. The data is taken from the Indian classical music dataset (Comp music dataset), which contains details of around 116 h. First, the features of audio such as Mel-Frequency Cepstral Coefficient (MFCC) and Mel-spectrogram are extracted. The extracted features given as input to the Bidirectional RNN model and the results are obtained with an accuracy of 75%.