Listening to Music in Groups

We evaluate the music taste representation computation and song matching capabilities of Jukefox, a smart music player for Android developed at the distributed computing group at ETH Zürich. We then use this music taste representation for single persons to create group playlists. A group playlist contains the aggregated songs of all group members, ordered from best to worst. The more group members like a song, the better it is for the group. We developed an application that uses the group playlist-concept to allow a group of people to automatically set up a listening session, where the application always plays the best songs for the group. The users can easily join the group by pressing a button, which connects the android phone to the server that runs a desktop client. The server communicates with the clients over (W)LAN to request their songs and music tastes. The server then computes the group playlist and requests the best songs from the clients. Once the server has received the mp3 files, it starts to play them. Clients can join and leave at any time; the server then automatically computes the playlist for the new group and, if necessary, requests new mp3 files.