Lorentz workshop on Music Similarity
19-23 January 2015
The Music Similarity workshop at the beginning of the year was my first workshop at the Lorentz center, and proved a week of inspiration in which some questions on what constitutes similarity in music were answered, but even more questions were raised. The workshop was organized by Anja Volk from Utrecht University, in collaboration with international music researchers Elaine Chew, Elizabeth Margulis and Christina Anagnostoupoulou.
Musical similarity can be approached from its audio data (encoded by mp3, wav and other file formats) and based on symbolic data (i.e. notation of music, as for instance scores of classical works, or transcriptions). Approaches from the audio or the symbolic domain provide their own chances and challenges; the organizers had therefore invited music researchers from many different fields, such as Musicology, Music Information Retrieval and Music Cognition.
In contrast to previous academic conferences I have attended, the Lorentz workshop format focusses on exchange and dicussion rather than individual talks. Every attendant is provided with office space, and the various offices can be used to split up into groups and discuss shared research goals. Hence, every day was a mixture of keynotes, which set general themes of the day, and discussion groups, which focussed on aspects of music similarity, and reported the results back in a plenary session.
Researching melodic stability in folk songs, I work in the symbolic domain, with the transcriptions of the Dutch folk song database. Even though the researchers and research assistants of the Tunes & Tales project have a fairly good idea of where melodies vary between variants, and where they agree very well, it is hard to formalize these notions. I had the chance to show my approach to a suitable method on a poster to various researchers I have been in touch in over the past few years, some of whom I had not met in person previously to the workshop. So in that sense, the workshop assembled the “who is who” of my research field, and I could get very rich and varied feedback on my work.
Through the diversity of research questions and approaches I also could see how my research relates to the big picture, and that problems I find challenging are met by another set of challenges in other research areas. For instance, in a performance of a jazz standard, the difference between the notated version and the performed version might be considerable – through improvisation, meter and chord changes – while jazz listeners would still consider them highly similar.
It was wonderful to experience so many researchers working on the big puzzle of music similarity simultaneously, from different sides and angles. The workshop has resulted in a roadmap for combined research efforts; moreover, a special issue on Music Similarity in the Journal of New Music Research is going to be finalized in fall.