This repository trains a neural network for music genre classification. It's currently based on the GiantSteps key dataset (https://github.com/GiantSteps/giantsteps-key-dataset).
Install dependencies:
brew install ffmpeg
brew install sox
python3 -m pip install -r requirements.txt
Get the giantsteps key dataset:
git submodule update --init --recursive
cd giantsteps-key-dataset
./audio_dl.sh
./convert_audio.sh
cd ..
Create dataset for training:
python3 feature_extraction.py
Train network:
python3 train_model.py
If you want to speed up training on Apple Silicon, refer to: