MUSIC GENRE CLASSIFIER
======================
This is a deep-learning project, prepared for BUETUGCSE472 course.
PREPARATION
-----------
1. Clone this repo and cd into it.
2. Download the GTZAN dataset from kaggle.
3. Move the dataset to the proper location.
mkdir -p ./data
unzip ~/Downloads/GTZAN_data.zip -d data/
4. Rename the dataset.
mv data/GTZAN_data data/gtzan
5. Create and activate a virtualenv.
virtualenv create mlproj
source ./mlproj/bin/activate # Linux
"./mlproj/Scripts/activate" # Windows
6. Install required packages with the following commands
pip install -r requirements.txt
7. Extract the features.
python src/extract_features.py -m -s
python src/prep_train_test.py # Skip if train-test sets are already split
python src/build_model.py
8. Deactivate the virtualenv (or simply close the terminal)
Genre Prediction of Manually Curated Songs
------------------------------------------
To predict the genre of songs outside the GTZAN dataset, you need to extract the
mel-spectrogram features. See the file `src/prep_test_set.py` for inspiration.
Also see `music-genre-classification.ipynb` to see how the trained model can be
used to predict genres.