This project is a music recommendation service. It draws from the Million Song Dataset, and attempts to recommend songs based on a list of liked tracks from the user. The recommendation engine is a deep learning artificial neural network implemented with Keras and TensorFlow. The model is run using Flask to host middleware APIs, and the frontend is based on Skeleton.
Install dependencies with
pip install -r requirements.txt
To get the dataset, you can either run setup.sh or the following commands
mkdir data && cd data
wget http://static.echonest.com/millionsongsubset_full.tar.gz
tar -xvzf millionsongsubset_full.tar.gz
Alternatively, visit the Million Song Dataset's site to download the file and unzip it manually.
Training the neural network may take some trial and error based on the hardware available. Running
./lib/main.py will attempt to train the network using the full dataset and default hyperparameters.
Once the network is trained and a workable model is created, you can run the web app that recommends songs. First, move your best working model (and all the associated files) into ./model/working/. Then, to standup web app, first run
cd frontend
python -m http.server
This will serve the static site html pages
In another terminal window, run
python src/api.py
This will stand up the middleware api endpoints.
Then visit the static site at http://localhost:8000 and try submitting some songs for recommendation.