This project was submitted as a part of Statistical Machine Learning course at IIITD in Winter 2020.
10 genre music genre classification on the GTZAN dataset. Comparative and exploratory analysis of the pre-processing routines applied, feature extraction techniques and the models used for identification are enclosed
The ipynb file is well documented and is self-explanatory.
The Zip file named Features.zip contains 5 datasets that are created from GTZAN dataset.
In the ipynb file you can see that the Part 1 and Part 2 are used to create the above datasets, if you don't want to create them as it may take some time, you can provide path to any of the csvs present in this zip file to run the project.
IF you do not want to run the dataset creation part, you can directly skip to part 2 as shown in the jupyter notebook
Features folder contains 5 files: songdata1.csv, songdata3.csv, songdata5.csv, songdata10.csv, songdata30
default path set in the notebook is Features/songdata5.csv as that dataset is our proposed one and performs the best.
The report is included for analysis and results.