Collaborative Filtering Comes To Independent Music Makers: Find What You Like Through Other People Tastes With inDiscover

"inDiscover is a service designed to help bring independent music to new listeners. We appeal to both musicians and people interested in hearing fresh new music, by recommending songs based on our user's tastes. Songs are provided by independent artists from around the world in mp3 format (the artists themselves are responsible for hosting the mp3s). By using a technique called collaborative filtering, we are able to predict how much a given user would like a specific song based on how they have previously rated other songs. After a user rates a song, they are presented with 2 playlists; one of songs that we KNOW that they like, and one of songs we THINK they will like.
By having users rate along different categories (eg. how well would a song be to listen to while working out?) we are able to compose playlists for each of those categories (eg. a playlist of songs we think you would like to workout to)."