spotify-playlist-builder

This project was build during a hackathon organized by the KTH AI Society in approximately 10 hrs.

The goal was to build a program that could create a new playlist based on reasent listened music by the specified user.

configuration

Before running the program a few things need to be initialized:

After creating a project on Spotify's Developer portal both the client_id and client_secret can be found and should be configured in config.yml.
In Spotify's Developer portal the Redirect URI should be set according to the one specified in config.yml.

running the program

Run the program by executing python3 run main.py.

Program

The program consists of a few parts explained below.

scraper

The scraper retrieves data from two sources:

Spotify playlist data: Gathering all songs in spotify's public playlists.
User data: Gathering the set of songs the user listened to most recently.

For each song the following features were collected:

danceability
energy
key
loudness
mode
speechiness
acousticness
instrumentalness
liveness
valence
tempo
duration_ms

recommender

The recommender clusters tracks from spotify's public playlists and find songs that are in the same clusters (and close to ) as your recenlty listened tracks.

playlist builder

The playlist builder takes the recommended music and exports it to a playlist created for the specified user. If no name is given a random name will be generated (e.g. thirsty-firebrick-snake)