this project is an (ongoing) attempt to visualise all my liked spotify songs. generated plots available here.
scatter plot of all 727 songs
- downloaded the metadata of all liked songs
- used spotify audio features api to extract acoustic features like
'danceability', 'energy', 'key', 'loudness', 'mode', 'speechiness', 'acousticness', 'instrumentalness', 'liveness', 'valence',
and'tempo'
- used t-SNE for dimensionality reduction and plotly for user friendly visualisation
while embeddings are quite simple, some observations are interesting
bob dylan songs grouped together
electronic music? VU definitely being an anomaly
rap/hip-hop
- attempt to generate better embeddings (auto encoders?)
- model the song genre
- look for other relevant features