The Computational Study of a Musical Culture through Its Digital Traces*

From most musical cultures there are digital traces, digital artifacts, that can be processed and studied computationally, and this has been the focus of computational musicology already for several decades. This type of research requires clear formalizations and some simplifications, for example, by considering that a musical culture can be conceptualized as a system of interconnected entities. A musician, an instrument, a performance, and a melodic motif are examples of entities, and they are linked by various types of relations. We then need adequate digital traces of the entities, for example, a textual description can be a useful trace of a musician and a recording a trace for a performance. The analytical study of these entities and of their interactions is accomplished by processing the digital traces and by generating mathematical representations and models of them. A more ambitious goal, however, is to go beyond the study of individual artifacts and analyze the overall system of interconnected entities in order to model a musical culture as a whole. The reader might think that this is science fiction, and he or she might be right, but there is research trying to make advances in this direction. In this article I undertake an overview of the state-of-the-art related to this type of research, identifying current challenges, describing computational methodologies being developed, and summarizing musicologically relevant results of such research. In particular, I review the work done within CompMusic, a project in which my colleagues and I have developed audio signal processing, machine learning, and semantic web methodologies to study several musical cultures.

Introduction

Music is a complex human phenomenon that can be studied from many perspectives, and the field of musicology has greatly expanded its original scope by embracing new and diverse research disciplines and methodologies. Especially significant in the context of this article are the more analytical and experimental disciplines that are now part of musicological research. Anthropology, for example, is fundamental for ethnomusicology, psychology is a core for cognitive musicology, and computer science not only lies behind computational musicology, but also supports many other types of musicological studies.

When doing musicological research, we focus on a unit of analysis, and typically each methodology is adequate to study specific unit types. We study units such as musical artifacts, musical interactions, musicians, or musical communities. Not every methodological approach, however, is adequate for the study of every unit type. In computational musicology a common unit is a piece of music, whereas in ethnomusicology there is also emphasis on musical communities. In this article I propose to expand the research methodologies being used in computational musicology by adding the cultural context into our research, thus incorporating aspects of ethnomusicology. Computational musicologists should also study musical pieces together with the repertory to which they belong, a repertory that has a common style and is supported by a community. For that we use the concept of musical culture to identify our unit of study, considering that a musical culture is a stylistically coherent musical repertory together with the community of people producing and receiving it.

In order to study computationally a musical culture, computational musicologists use the available digital traces. There is an exponential growth of available digital data of relevance for computational analysis, and especially on the World Wide Web (WWW) we have access to large quantities of music-related data. In order to perform music research, however, the data have to be adequate for the problem to be addressed, and most available data are not suitable for most musically relevant problems. We use the terms, corpus and datasets, to refer to the types of data collections that are needed. A music corpus is a large collection of structured data that can be used to represent a given musical repertory, and datasets refer to the specific collections of structured data, especially including expert annotation data, that are used to perform specific experiments.

The process used to perform computational studies of a musical culture can be represented by the diagram in figure 1. As it is shown, we first gather data from the WWW (or any...