Music Genre Classification with CNN Model Evaluation

Yoonhee Jang · 2023

Building a music genre classification model is one of the steps to consider the copyright. Through tuning, I was able to find the reasonable setting for the model. CNN shows high accuracy results rather than other Deep Neural Networks. It shows better results when the validation split is small and the filter size is bigger than 32. Additionally, batch size gives partial impact on the accuracy. Epochs can be the main factor to control the result but it is not for the music data.

Read the paper · More papers on PaperTik