Perspectives for Computational Musicology

This symposium was held on 5 October 2010. It was organised by the Meertens Institute and the Department of Information and Computing Sciences (Utrecht University) on the occasion of Peter van Kranenburg's PhD defense on 4 October in Utrecht and the conclusion of the CATCH project WITCHCRAFT.


This symposium was kindly supported by a grant from the Netherlands Organisation for Scientific Research NWO.


9.30 registration, coffee, tea
10.00 Opening by Theo Mulder, scientific director of the KNAW
10.15 Peter van Kranenburg: "Introduction: Doing Computational Musicology", abstract; presentation.
10.45 Tim Crawford: "Is Computational Musicology a distinct discipline, and should it be?"
11.15 Anja Volk: "A moment of opportunity for Computational Musicology"
11.45 Henkjan Honing: "Surprise! Assessing the value of risky predictions", abstract; presentation.
12.15 Lunch
13.30 Christina Anagnostopoulou: "Can computational music analysis be both musical and computational?", abstract; presentation.
14.00 Geraint Wiggins: "Understanding the experience of musical listening"
14.30 Coffee, tea
15.00 Olmo Cornelis: "TARSOS, platform for pitch analysis of ethnic music", abstract; presentation.
15.30 Theodor Dumitrescu: "The Text-critical Perspective"
16.00 Frans Wiering: catch-up and general discussion, presentation.
16.30 drinks


Peter van Kranenburg (Meertens Institute): "Introduction: Doing Computational Musicology"

This introductory talk proposes an abstract model for "doing" computational musicology. What is essential in computational modeling of musical phenomena? In order to design a meaningful computational model, interpretable relations between the model and the music it descibes are of crucial importance. Therefore, designing a computational approach to study a particular musicological problem involves at least: 1. understanding the musical problem, 2. designing a proper datastructure and algorithm, 3. musical interpretation of results, and 4. revision of the model. Some examples in which well-known algorithms from computer science are used to study music will be presented.

Tim Crawford (University of London, Goldsmiths College): "Is Computational Musicology a distinct discipline, and should it be?"

Musicology is by its nature somewhat multi-disciplinary, mainly because there is no single - uni-disciplinary - mode in which music can be analysed or discussed. Music is not simply its sound, or its notation, or the social/historical/political context of its composition/performance, or its emotional impact, or whatever. Such features might each independently provide a handle on an aspect of music - a separate mode of understanding of it - but not everything about it. Perhaps computer tools really can offer some new modes of understanding, but these will need interpreters to translate them into language that 'traditional' musicologists can understand. For many years I have been an optimist about what computers can do for the musicologist, but, to be honest, little of what I hoped for when I started has yet come about. In this talk I shall draw on my own experience to illustrate some of the ways in which I still hope that the use of computers can supplement my own musicology. It is hard to escape a nagging feeling that it would in some sense be a failure if such work were to be labelled as 'Computational Musicology', implying that it took place in its own isolated domain or sub-discipline.

Anja Volk (Utrecht University, Computer Science): "A moment of opportunity for Computational Musicology"

Nicholas Cook stated in a keynote lecture at ISMIR 2005 that we are standing at a moment of opportunity regarding the relationship between computational modeling and musicology. This talk discusses chances, challenges and obstacles of computational applications in musicology in the past and reflects about current challenges that have to be met in order to realize the opportunity.

Henkjan Honing (University of Amsterdam, Musicology): "Surprise! Assessing the value of risky predictions"

The eventual objective of this research project in preparation is to analyse and quantify the role of surprise in the confirmation of computational models of music cognition, taking into account the surprise following unexpected empirical findings, as well as the surprise stemming from unforeseen empirical consequences of the models. The project presents a challenging case study to confirmation theory, aimed at improving statistical methods in the computational humanities and in cognitive sciences. At the same time, the project will resolve some pressing questions on model selection in theoretical musicology and music cognition.

Christina Anagnostopoulou (University of Athens, Faculty of Music Studies): "Can computational music analysis be both musical and computational?"

The area of computational music analysis has been developing since as early as the 60s. Yet, there are no universal fully automated analytical systems to date, meaning that the process has so far resisted full formalisation. The issue therefore remains. This talk will look at several key points in the area of formal music analysis, such as the choice of analytical criteria, segmentation and categorisation, and discuss ways these are treated using computational methodologies. Emphasis will be given on the evaluation of the obtained results, especially in cases where statistical significance is involved.

Geraint Wiggins (University of London, Goldsmiths College): "Understanding the experience of musical listening"

One aim of music analysis, in contributing to musicology as a whole, is to explicate the experience of musical listening, as experienced by an archetypal listener. With the advent of computational cognitive models of musical listening, it becomes possible to simulate this process automatically; when those models are neurophysiologically validated, a claim can be made that the predictions of the model are related to the human listening experience. This talk concerns the application of such an approach to Syrinx, by Debussy, and presents preliminary evidence that automated, cognitively-motivated analysis can usefully contribute to the study of music.

Olmo Cornelis (University College Ghent, Faculty of Music): "TARSOS, platform for pitch analysis of ethnic music"

Ethnic music is a vulnerable cultural heritage that has received only recently more attention within the Music Information Retrieval community. However, access to ethnic music remains problematic, as this music does not always correspond to the Western concepts of music and metadata that underlie the currently available content-based methods. During this lecture, we like to present our current research on pitch analysis of African music. TARSOS, a platform for analysis, will be presented as a powerful tool that can describe and compare scales with great detail.

Theodor Dumitrescu (Utrecht University, Musicology): "The Text-critical Perspective"

The field of musicology has lived through decades of ongoing labor in bringing early repertories to light through the creation of modern editions, a formidable activity leading to not only commercial printed scores but also a tremendous host of doctoral dissertations and other academic studies. These have not always been the most critically engaged examples of scholarship, but they continue to serve as our primary portal to the music of the past. In the computing world, the situation with musical materials is more extreme: the notated music used as raw data typically escapes critical scrutiny altogether; the fact that the data exists at all is implicitly seen as sufficient.
This is a problematic attitude. Beyond the issues with basing research on faulty or deprecated material, musical data by its nature resists encapsulation in a single definitive form. The transmission of music is prone to variability, and this remains true from medieval repertories on even to the modern day. Many of the problems are common in literary transmission, but some are also especially pronounced in music: Text-critical methodologies, developed continually since the 19th century, offer a starting point for helping us deal with the often complex problems of transmission and variability, even if the methods themselves are in dire need of revision. Conversely, computational techniques can aid musicologists and performers in developing these revisions, by opening their materials to large-scale algorithmic operations and, quite importantly, permitting multiple methods to go through quick development cycles. The price of all this is that the musical and variant data must be there, and not simply any data; in order to be effective it needs to be created and maintained under the supervision of experts in the appropriate repertories.