Automated Music Classification using Machine Learning for Indian Songs
Kiran Puttegowda, Kay Hooi Keoy, R Deepak, Vinaye Armoogum, B D Parameshachari · 2024
Automated classification of music-genre is a very active research topic, as it is significant in music information retrieval system. There is lots of research done in classifying western music genre. India has a heritage of rich music culture, from Purana’s, Veda’s, to modern times, music is an essential part of Indian households. In India, Music genre differs with the change in geographical area because of different cultures, religions, languages etc. This makes automating human classification abilities for classifying various music-genre a hard task. There is also very few data set to work on, for Indian Music Classification. In this work, we are releasing a database for Indian songs genre (Bollywood Romantic, Rap, Bhajan, Garhwali, Ghazal, Bhojpuri and Sufi) classification and compare the performance of different ML algorithms (KNN, Neural Network, Gradient Boosting, SVM and Light GBM) for the classification task. Light GBM classifier gives the best accuracy of $87.2 \%$.