This is a visualization created in processing based on the contents of me and friends’ music libraries. I gathered the data from each person through an XML export of their iTunes library and then created this visual tool in order to help us figure out who has what music, what music we might not have heard about and just to marvel at the sheer amount of music we collectively own.
This is the first iteration of a project that creates a different way to compare music collections with your friends. I emailed my friends and asked them to export their iTunes libraries and send them to me. I parsed the data keeping track of each band, album and track. For each track, there is a list of people who currently own the track. Each group of blocks corresponds to a band and the size of the block indicates the total number of tracks shared between friends. The color of the blocks indicates the person who owns the band. If more than one person has a band in their collection, the blocks are connected with a line. When you click on a band, you see all the albums with the tracks color coded so you can see who currently owns the track. The objective is two-fold: 1. See all of the music your friends are listening to that you may not have heard of and 2. See if there is someone you know who may have the missing album that could complete your collection. I wanted to create a different experience than just scrolling through a list of bands, as if you were looking through their library on their computer. I wanted it to be a random experience, something akin to record shopping. What I love about record shopping is the chance. You never know what treasure you are going to find buried in the bins. So, when accessing the bands, I kept it chaotic and random but when you look at the albums, it’s more straight forward. If you are looking for a band in particular, you can use the search function to filter the results. Now for the nitty gritty, for all interested parties, I built the program in processing and used both the controlP5 and PeasyCam libraries.
I am currently researching audio extraction algorithms to use for the second iteration of the project. Currently, locations are assigned randomly but I plan on assigning a location based on a self organizing algorithm on the bases of audio features extracted from each song. That way, similar sounding songs, artists or albums (depending on view) will be located near each other. In order to refine the algorithm, I am also creating an “admin” view which will allow me to assign a particular weight to each value in order to tune the algorithm so I can visually observe how it is comparing songs. This will be helpful because the organizing algorithms efficiency can vary largely with the type of audio that it is trying to compare. I should have something for everyone to look at and play with by the beginning of December.