Research Paper Link: https://github.com/AndyLiuCodes/Music-Genre-Classification-with-CNN-Hybrid-Architectures/blob/master/Music_Genre_Classification_Report.pdf
Instructions to Generate Data: Have the /fma_small folder in the main folder of the repo Select the number of files to generate in data_setup.py run python data_setup.py and all the folders and file will be generated
To train the model: specify parameters in load_data_generators.py specify model in train.py - specify number of epochs, etc. specify parameters in model.py - this is the file that contains all four models run the command "python train.py"
outputs your final model, best_model in .h5 format
To test the model go to predict.py and specify which model to test, either model.h5, or best_model.h5 run "python predict.py"
this outputs your final test accuracy and loss in the command line
Make sure you have the .gitignore file so that the actual data does not get pushed onto the repo
Contributers: Andy Liu Henry Yip