MIAS

A Spotify song recommendation system that allows weighting of acoustic features that are important to you, giving you in-tune song recommendations.

The initial idea behind this project is to built a recommendation system to help artists expands their performance playlists, by identifying similar songs.

The hosted and running application can be found at: https://mias-recommender-sys.streamlit.app/

Developer

Please make use of the Developer documentation available here: https://trav-d13.github.io/MIAS/.
The most upt-to-date dataset is openly available from: https://mias-recommender-sys.streamlit.app/dataset
Alternatively, the dataset is hosted on Kaggle and available at: https://www.kaggle.com/datasets/travisdaws/spotify-tracks-acoustic-features/data

Disclaimer: The dataset was collected to create a simple song recommendation system, based on track acoustic features, artist, popularity, etc. It does not include any propriety Spotify audio content or Spotify audio tracks. The dataset contains publicly available information from Spotify and is intended only for learning. The application deployed is only for testing creating a usable recommender system, it is not intended as a product for commercial or intensive use. Please respect it, and the Spotify Developer API.

Copyright: Shield:

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.

Project Structure

MIAS/
│
├── .streamlit/
│   └── secrets.toml
│
├── data/
│
├── notebooks/
│
├── docs/
│
└── src/

Running the Application

Docker Container

This section deals with building and running the docker container after making changes in the project. Please note, that changes to code will not require the Docker image to be re-build, but altering the requirements.txt, Dockerfile, etc., will require the image to be rebuilt.

Build the Docker image

sudo docker build -t mias .

Run the application

sudo docker run -p 8501:8501 -v ./data:/app/data mias

The container allows for a volume containing all the required data to be linked to the container to allow data access.

Manual Running

streamlit run src/app.py

Spotify Developer Credentials

The application makes use of Streamlit's secrets manager. For the running applications these are stored in the Streamlit running platform. In order to make the secrets available during development, please perform the following steps:

Application

Find both your Client ID and Client Secret from Spotify Developer.
Create the following file .streamlit/secrets.toml from the project root directory.
Fill in the following information in the secrets.toml file

CLIENT_ID = "CLIENT ID KEY"
CLIENT_SECRET = "CLIENT SECRET KEY"

The application now has access to your spotify credentials.

Notebooks

The notebooks make use of environmental variables, please make sure you have a Python file named notebooks/credentials.py. Inside of credentials.py please insert the following with your correct Spotify configuration keys.

import os

def set_credentials():
    os.environ['CLIENT_ID'] = ""
    os.environ['CLIENT_SECRET'] = ""

Spotify Data Collection

Automated Collection

The currently running application runs a Chron jobs to scrape and collect tracks from the top-performing playlists every two-weeks, updating the tracks dataset. Additionally, every playlist queried for recommendations is added to the dataset. The dataset is freely available, please see the bottom of the dataset page on the application available at: https://mias-recommender-sys.streamlit.app/dataset or alternatively the Kaggle dataset at: https://www.kaggle.com/datasets/travisdaws/spotify-tracks-acoustic-features

Manual Collection

If you would like to test the data processing and manually collect target playlist or general spotify data, copy and paste the below script to the bottom of data_processing.py and execute it using the following command in terminal: python src/data_processing.py

if __name__ == "__main__":
    client_credentials_manager = SpotifyClientCredentials(client_id=st.secrets['CLIENT_ID'],
                                                          client_secret=st.secrets[
                                                              'CLIENT_SECRET'])
    sp = spotipy.Spotify(client_credentials_manager=client_credentials_manager)

    url = "https://open.spotify.com/playlist/799B2k7VQhsWeA2iQrun9f?si=345d3d94fb484f2c"

    # target_playlist_extraction(sp, url, "Rob Performance Playlist")
    top_playlist_extraction(sp)

trav-d13/MIAS