A Music Library in the Palm of your Hand: A Series of Experiments on Interface Culture

We are interested in the socio-cultural compatibility of media-related technology, especially music information retrieval (MIR) systems. In order to study this special topic we have undertaken several attempts in the past to design and implement applications and experimental scenarios. One solution to observe users’ interactions with large-scale music libraries outside the lab was a mobile application, the so-called ‘musicAssistant’. It allows users to directly interact with a server-side music database over wireless LAN. An integrated recommendation engine may be influenced by a virtual joystick, which controls the weights of 3 musical facets (sound, cultural and lyrics) each contributing to a combined music similarity measure. We collected data with this framework and evaluated the relations between similarity facets, namely sound, lyrics, and cultural issues. Additionally, we have been going beyond the idea of pure similarity-based recommendations by integrating third-parties’ metadata schemes and instances in order to re-use voluntary and commercial efforts in building music ontologies. Open application programming interfaces (API) of commercial players such as Google and Amazon allowed us to concentrate on the implementation of mobile user interfaces rather than collecting metadata on our own. In a second attempt we exhibited stationary MIR systems during a public event and drew implications of its outcome for our future work. In this paper we want to sketch first the basic aspects when implementing systems for personal music libraries. Then we will describe the architecture of such mobile and stationary music libraries and the involved MIR technologies (multimodal similarity, self organizing maps, interface metapher). We report about our initial findings in different scenarios and close with a sketch for future users’ needs.