What is this for?

The Spotify API has an audio analysis endpoint that gives you lots of useful info about a track. Among other things, it gives you its AI prediction of what key and mode a track is in: e.g. C major, A minor, G# major, etc.

This lets you compile playlists based on key & mode, which is useful for people learning a musical instrument. Let's say you're currently learning the A major scale on the piano: you could take an existing playlist filled with songs you'd like to learn to play and filter them based on A major, compiling the results into a custom playlist. You could also compile a playlist filled with ambient songs to use as backing tracks, filtering them based on key & mode.

How do I use this?

• This script requires Spotify client credentials (see instructions here). The .env file at the project root expects credentials and a redirect uri. Insert them.
• Then, run python main.py. This will trigger a setup script when run for the first time. The setup will involve authenticating yourself into your own Spotify application, giving it access to your account.
• Then, add some playlist IDs (newline-separated) into playlist_ids.txt.
• Finally, rerun python main.py. This time, you'll be prompted to select a key & mode, as well as tick the playlists you'd like to compile from (tick using whitespace). The newly compiled playlists will appear in your Spotify library.

Caveats

• The compiled playlists are public by default. If you want to make them private, do it in the Spotify UI itself.
• If you compile from the same playlist more than once, the already-existing compilation will be updated instead of creating a new one.
• If you delete a track from a compiled playlist, it won't be re-added if you run the same compilation script again.
• If you manually add a track to a compiled playlist, it won't disappear after you run the same compilation script again.
• If you delete a track from the original playlist you're compiling from, it won't be removed from the compiled playlist. The logic is that you might be using auto-updating playlists like Discover Weekly, in which tracks are regularly removed automatically to be replaced with new ones.
• If you delete a compiled playlist, all its track associations will be forgotten, and you'll be able to rerun the same compilation script again from scratch.
• The tracks in the compiled playlists are sorted based on tempo in ascending order (i.e. the slower tracks are at the top, faster at the bottom).