Frederick O'Loughlin

Spotify Song Data Summary

Playing with the Spotify API

How do you quantify a song? It turns out it's really difficult. That hasn't stopped us creating thousands of music genres to loosely cobble together songs. Such groupings tend to be as much about the time and place the song was created then the music itself.

If then you wanted to recommend a song based on the music itself rather then the culture it was created from. How would you do it?

This is a challenge for data science and one that Spotify has risen to. Unknown to most listeners, Spotify has categorised songs by a variety of features including daneability and energy. The fact that Spotify have been able to perform such feature extraction automatically on their library of songs does in large part explain their powerful recomendation services.

I enjoyed looking deeper into some of this by playing with the Spotify API. I built a program that automatically mapped out all of the information visually for a song.

I have decided to not build this into a service yet. But let me know if you think I should.

(Photo credit Lee Campbell, via Unsplash)

Read Spotify's API Documentation