Pengembangan Aplikasi Klasifikasi Musik Reggae

Telyoarubun, Peter Hendriks

KALBISIANA Jurnal Sains, Bisnis dan Teknologi
Vol. 10 No. 3 (2024): Kalbisiana

This study aims to design software to classify Reggae music and other genres of music. The training data used is music audio data totaling 721 music audio and 309 audio data for test data. The method used in this study uses Scikit-learn software to create machine learning models and uses Flask software to create application displays. The final result in increment one is the accuracy of the training data and test data at 150 epochs, for the results of the training data accuracy of 75% and test data of 65% and getting a model evaluation with the results of Accuracy Score 95.27%, Precision Score 95.99%, Recall Score 98.87 %, F1 Score 97.41%. The results in the second increment are to produce a website-based application to select music, play music and get predictive results in the form of Reggae and Not Reggae.