Classification of Indian Classical Carnatic Music Based on Raga Using Deep Learning
S John, M S Sinith, R Sudheesh, P. P. Lalu
Abstract
Audio signal processing can be integrated with music domain in-order to retrieve contents within the music signal. Raga is often related to melodic framework in Indian classical music. Automatic Raga identification can be achieved through audio signal processing and Raga identification can be used for different applications including music signal synthesis, audio dataset management and music therapy. Classical methods for Raga analysis in time and frequency domain are less efficient and these algorithms are quite time consuming. The overall accuracy of Raga classification can be increased by incorporating artificial neural networks. In this paper, pitch contour of selected Ragas in Carnatic music are used as key features and model implementation is done using Convolutional Neural Network (CNN).