With this program, witten in python, you can import your spotify playlist and it will create one for you.(That is the goal) Right now it is mainly used as a data gathering app, then that data will be used to created the playlist. Spotify does a good job of creating playlists for their users, however there are some areas I hope to improve upon.
If you would like to run this program you will need to do some legwork before hand since I have yet to build a solid UI. Go Here:https://developer.spotify.com/documentation/general/guides/authorization-guide/ and create a spotify app, to get your personla identification to use. Then export those stats by following these directions: https://spotipy.readthedocs.io/en/latest/#authorization-code-flow
It would also be good to have your Spotify user-Id(not your name) at hand.
There is a lot of data anylisis right now, becuase I would like to make sure that I get a good understanding about the data before I start using it to create the playlists.
Make Playlist:
python3 main.py Playlist
Delete Playlist with playlist id:(Used for unit testing when creating multiple playlists in a row)
python3 main.py Playlist playlist-id-here
Get some data on songs and such:
python3 main.py GetData
I heavily rely on the spotipy: https://spotipy.readthedocs.io/en/latest/ library, and its great. I have been told that updates stopped about a year ago, however it still works great!
Loudness throughout a song is important when making a playlist becuase, you can make it so that the user has one song that begins with the same loudness as one song ends, for a better listening experience.
Here are four graphs that represent 4 different recomendation requests. What we are looking at is how similar are the features of the song "seeds" to the songs that spotify recomends. The higher the feature on the graph the less likely it is that recomendations uses that feature of a song. The closer to 1 that the feature is the more likey that it is used. Some are labled for observation. I used my 5 top tracks, and then asked for recomendations based on those 5 tracks.
We can see that Danceability, Valence and Instrumentalness are usually very close to one, therfore they may be used more in recomendations. While Tempo and Key are usually quite high/different, and are therefore likely less used when making recomendations.
Recomendations is how spotify gets new songs for users. This is usefull in a playlist becuase if a person is partial to a specific key, or tempo of a song, then they may want their playlist to reflect that. However right now it seems that spotify does not take those features into account. While those features may not be as important as valence or energy, I think it should be taken into account. It is possible to add more inputs into the recommendations request, this analysis was done by using 5 songs.
Decide stats to use on making playlist.
Want solid data analysis
can do make type of playlist: Sad, Happy, Workout, Study
LOOK INTO KEY AND MODE on the effects on mood for more analysis
Workout tracks should have: high energy, High danceability, High Energy, high loudness
Study/Chill tracks should have: High Valence, low tempo, low loudness, high instrumentalness
Sad tracks should have: Low Valence, low tempo, low loudness, low energy
Happy tracks should have: High Valence, High liveness (Key and mode need to be researched more)
If someone wants a playlist that takes a certain amount of time to finish
Get solid UI done, with stats display and such.(With Azure)
Integrate with twitter, if person follows the artist.
Hans Zimmer: dunkirk, Interstellar, Sherlock Holmes, Momento : Shepard tone. Shephard Risset-Glissando
The prestige by david julian