Music mood taxonomy generation and classification of Christian Kokborok song: an audio-based approach.

Music information retrieval (MIR) is a growing field of research and mood classification is one of the applications of MIR which represents the relationship between human emotion and music. In recent decades, it is performed on Western and Indian languages like Hindi, Telegu. We are working on an under-resourced language like Kokborok. Kokborok is a regional language of the northeastern states of India and spoken in countries like Bangladesh, Myanmar, and Bhutan etc. We have developed an audio-based system for classifying modes of Kokborok song using prominent features like rhythm, intensity, and timbre. Decision tree classifier j48 is used for the classification purpose. Our dataset composed of 125 songs and clips of 30 second is being used to build the computational model that consists of four different mood clusters, and each cluster has three subclasses. We have achieved 54.4 % accuracy rate for music mood classification on the above data.