We examine the use of computers in storing and manipulating music. We consider the validity of treating music as information in the formal terms required by computers. We take the metaphor of inscriptions (marks on a medium) and draw out its implications for music representation techniques and digital encoding practices through its relationship to notations and to digital storage, its ability to take on semantics and become a representation, its ability to be gathered together into documents, and its ability to be disseminated, particularly over digital networks. We then examine some examples of practice in designing and applying digital music encoding methods and draw some conclusions for the practice of computer assisted musicology: that suitable encoding methods are vital for any application of computers in music research, and that users must understand how musical information is being represented in order to make optimal use of the techniques.

Transforming Musicology is a three-year project undertaking musicological research exploring state-of-the-art computational methods in the areas of early modern vocal and instrumental music (mostly for lute), Wagner's use of leitmotifs, and music as represented in the social media. An essential component of the work involves devising a semantic infrastructure which allows research data, results and methods to be published in a form that enables others to incorporate the research into their own discourse. This includes ways of capturing the processes of musicology in the form of 'workflows'; in principle, these allow the processes to be repeated systematically using improved data, or on newly discovered sources as they emerge. A large part of the effort of Transforming Musicology (as with any digital research) is concerned with data preparation, which in the early music case described here means dealing with the outputs of optical music recognition software, which inevitably contain errors. This report describes in outline the process of correction and some of the web-based software which has been designed to make this as easy as possible for the musicologist.

Funded under the AHRC/EPSRC/JISC e-Science in the Arts and Humanities Initiative, the e-Science for Musicology workshop held in Edinburgh in July 2009 brought together expert speakers in using computer technology in music scholarship and musicologist participants with little or no experience in such methods. During the course of the workshop, several techniques for dealing with music in computers were demonstrated and the participants were given the opportunity to comment on those techniques and discuss their wider implications for the discipline.

Music Information Retrieval has been an active research area for over a decade and much is now known about how to search music scores in representations such as MEI or MusicXML. However, this must usually be done by composing a query in a specialised search language. On the other hand, there is a vast body of musicological literature containing detailed analyses of musical works. These texts make frequent references to musical passages in scores using natural language phrases which do not relate directly to search expressions. Our long-term aim is to investigate whether these phrases can be linked automatically to the musical passages to which they refer. As a first step, we have organised for two years running a shared evaluation task in which participants must develop software to identify passages in a MusicXML score based on a short noun phrase in English. In this paper, we present the rationale for this work, discuss the kind of references to musical passages which can occur in actual scholarly texts, describe the first two years of the evaluation and finally appraise the results to establish what progress we have made towards our goal.

Performance of a musical work potentially to provides a rich source of multimedia material for future investigation, both for musicologists’ study of reception and perception, and in improvement of computational methods applied to its analysis. This is particularly true of music theatre, where a traditional recording cannot sufficiently capture the ephemeral phenomena unique to each staging. In this paper we introduce a toolkit developed with, and used by, a musicologist throughout a complete multi-day production of Richard Wagner’s Der Ring des Nibelungen. The toolkit is centred on a tablet-based score interface through which the scholar makes notes on the scenic setting of the performance as it unfolds, supplemented by a variety of digital data gathered to structure and index the annotations. We report on our experience developing a system suitable for real-time use by the musicologist, structuring the data for reuse and further investigation using semantic web technologies, and of the practical challenges and compromises of fieldwork within a working theatre. Finally we consider the utility of our tooling from both a user perspective and an through an initial quantitative investigation of the data gathered.

The AHRC-funded Transforming Musicology project includes a work package studying the leitmotive technique of Richard Wagner, used in his operas to associate characters and other key plot features with musical themes. We are approaching this from three angles: examination of the extensive literature on the technique; psychological studies of listeners experiencing the music; and audio search within collections of recordings of the operas. The musicological literature includes various claims that the leitmotives function partly by eliciting emotional responses in listeners. We report here on the data capture phase of a large scale experiment in the psychological domain of this work which attempts to examine these claims.

Wagner's cycle of four operas, Der Ring des Nibelungen, is one of the best sources of the leitmotive technique in his output. The complete cycle was performed at the Birmingham Hippodrome by the Mariinsky Theatre in November 2014. We funded free tickets for ten students from the Birmingham Conservatoire to attend all four operas and participate in our experiment. Each participant was fitted with a device manufactured by a company called Shimmer which measures their galvanic skin response, heart-rate, and movements. Electrodermal activity (EDA) is a measure of sympathetic and parasympathetic activity of the autonomous nervous system and is indicative of emotional arousal (e.g. Boucsein 2012). Several studies have found EDA and skin conductance responses to be a very sensitive measure of music-induced emotions (e.g. Grewe 2007; Khalfa 2002; MasHerrero 2014). Along with the biosignal data, we also captured the sound audible from the participants' location in the auditorium (which we will align with a commercial recording of the operas for presentation purposes) and detailed, time-based expert musicological responses to all aspects of the production (see Nurmikko-Fuller et al. for a detailed description of this aspect of the work).

These three components (sound, biosignals, and expert responses) result in a rich, multi-modal representation of a musical performance. We describe the process by which we were able to time-align the data streams that make up this resource. We describe the software infrastructure we are using to allow us to identify emotional arousal in the biosignal data. We make a case for the relative musicological objectivity of our data and compare it to other forms of music criticism. We describe our plans for future Linked Data publication of this resource to allow other researchers to test hypotheses about emotional responses to music.

Under its Digital Transformations in the Arts and Humanities theme, the Arts and Humanities Research Council recently awarded three large grants to projects which will explore how digital technologies are transforming arts and humanities disciplines. Transforming Musicology, led by Tim Crawford at Goldsmiths' College and with partners at the Oxford e-Research Centre, the Oxford Music Faculty, Queen Mary University of London, and Lancaster University, has been awarded funding under this scheme. The project presents an opportunity to transform musicology's impact and status by recognising the full implications of its multidisciplinary nature and taking advantage of the potentials of technology for studying the vast musical resources of the internet, both those already available and those which may become available in future.

The world of music has been already transformed by the digital revolution. The same technology that has given unprecedented access to music for a vast and ever-growing international audience can open musicology up to the world---anyone and everyone can contribute by a variety of means enabled by that technology. We will show how the computational tools of music information retrieval (MIR) can be enhanced and adapted to the needs of musicologists, and how state-of-the-art developments in the Semantic Web can be exploited both to make their work more relevant and more sustainable, so that their methods can be easily re-used on new data. We will enhance and extend Semantic Web technology for making richly multi-modal music available on the Web and explore Semantic Web techniques for capturing and analysing musical discourse---both scholarly and popular---around such music collections. The work will draw on the extensive experience of the Oxford e-Research Centre in digitally representing research processes and apply it to the inherently humanisitc methods of musicology.

This project will stimulate creativity through multidisciplinary collaborative working. In the past, musicologists have tended to be lone scholars; regular collaboration in a multidisciplinary research environment will in future be essential for them to find what is ``interesting'' in potentially huge collections of music.

We shall be using MIR tools in three main strands of research. Two of these are typical subjects for musicology and will be conducted by world-leading experts: 16th-century music and Wagnerian leitmotives. These will use state-of-the-art score- and audio-analysis techniques to extend the range of musicological investigation beyond the normal limitation to printed scores. We will also be doing psychological work at Goldsmiths to study why leitmotives are so recognisable (or not). Our third main research strand is entirely novel, being nothing less than a prototype for a new ``musicology of the social media''. Music is now created, recorded, distributed, re-used and shared entirely online throughout the world and we will explore a whole new way of studying it and the online spread of musical ideas.

Engagement with the communities of practice who are potential clients of MIR is key to the success of Transforming Musicology. Therefore, we plan to solicit a small number of funded mini-projects in which musicologists and technologists will collaborate for six months on a research idea of their own devising. The outcomes of these projects will be show-cased in both presentation and published form and will feed into our understanding of how well musicologists can adopt digital techniques.

As well as musicology, another crucial community of practice is the recorded music industry. Through this project technologists in that industry will see that musicology helps them answer the needs of their clients by offering better means of access to the music they want to hear. The digital revolution in music delivery and re-use raises many crucial issues to do with creativity, authorship, marketing, copyright and commercial value beyond those normally associated with musicology. Combining the analytical methods of musicology and the power of digital methods brings intellectual rigour and empirical validity to these debates. Musicologists need this cross-disciplinary background to contribute most effectively and to create new and valuable roles within the music industry.

Musicology should not be an "ivory-tower" discipline; most people are interested in where music comes from, how it is conceived and made, how it affects us and society. Music goes on in people's minds, and the study of music is essentially about interactions between people. Because of this, music raises a complex mixture of philosophical, psychological and intellectual challenges; so it is a particularly fruitful domain for working through transformative technical challenges which will be more widely applicable to the Digital Humanities and beyond.

Numerous tools and techniques for working with music and sound to a level which could contribute significantly to music scholarship have been developed in recent years, yet they have not so far influenced mainstream musicology. This paper seeks to address the question of why this has not happened.

The work comprises some specific case studies of musicologists examining: what their scholarly practice consists of; what kinds of computational techniques they already employ; what specialist sound/music computational tools they are aware of and use; their opinions on the issues arising from computation in the study of music, especially the idea of music as data and the operationalisation of musicological practices.

The case studies are being carried out over the summer and autumn of 2011 and will comprise semi-structured interviews, observed interaction with computational tools, and knowledge elicitation tasks. The data gathered will help to establish a grouned understanding of why music scholars have not been exploiting computational techniques to their full potential.

Electronic resources facilitate the exploration of materials in ways that are otherwise difficult to achieve. Where often such resources are conceived as corpora in the form of collections of objects - for musical corpora, usually musical works - and indexed by `metadata' - resembling library catalogue entries - it is not clear that this is appropriate for a discipline that is rarely concerned solely with fixed, purely musical material. The Electronic Corpus of Lute Music (www.ecolm.org) was designed to provide substantial historical and bibliographical context, along with images and text-critically annotated musical materials.

With ECOLM, we intended to show the potential electronic resources have to effect dramatic changes to the way in which information is disseminated. Such resources allow the juxtaposition of materials that are difficult to otherwise combine, permit scholars to contribute small observations - too small to publish individually - record the provenance of annotations and of the materials added and permit levels of access that would previously have been geographically or financially limiting especially for primary materials.

By providing provenance, they also allow users to become better informed and active critics of the information they receive, with multiple interpretations more easily evaluated side by side. By providing ways of managing large numbers of links and connections, they can allow a substantial degree of contextualisation, the exploration of which can be driven by the interests of the user rather than the compiler. By permitting user-contributed content, a project can outlive its funded time and provides a cost-effective and direct means for a research community to collaborate in the sharing and dissemination of the content and the new knowledge derived from it.

Such a resource does raise issues, including the status of such publications, quality control, copyright and intellectual property, and managing the quantity and diversity of information that can accumulate. In this paper, we discuss these and consider the impact such resources will have on the discipline. The presentation will draw on our experiences with both ECOLM and the Purcell Plus project.

For the last half century or so, musicology has been a discipline whose practitioners, when they engage with musical works, have, for the most part, been concerned with close reading of those works. From its inception as a kind of musical philology in the nineteenth century up until the second world war, musicology was much more disposed to embrace comparative techniques; music was assumed to embody universals which could be discovered by stylistic comparison of the works of newly formed canon or by comparison of musical practices from numerous cultures. After the second world war, however, this belief in musical universals was largely rejected. Stylistic comparison was replaced with musical analysis which held that the interesting features of musical works were unique to those works and that comparison between works was not of interest. Comparison of cultural practices was replaced by ethnomusicology which held that the musical practice of a culture was unique to that culture and the comparison between cultures was not valid. So we inherit a discipline which looks for interest in the minutiae of musical details, or in distinctive cultural practice.

In the last ten years, however, musicology has begun to revisit some comparative practices in the form of empirical musicology, doing musicology with evidence. Generally this has meant (for the small number of scholars engaging in these practices) a shift of focus away from exceptional details and towards examining large quantities of musical information and focusing on statistical trends. This re-orientation within a small section of the discipline has meant taking on some important new conceptions of music and of musicology; that of music as information (as opposed to---but not replacing---the nineteenth century view of music as art work or the twentieth century view of music as practice), and that of interdisciplinary collaboration (conceding some of the work, especially technical aspects of the discipline, to scholars from other disciplines and to technology). So we are now faced with a discipline which takes seriously the possibilities of comparative research and computational discovery.

Enter e-Science. While the importance of enhancing or even enabling research methods by the use of (particularly computer) technology is a vital aspect of e-Science, Paul Wouters (2006) describes two important new classes of "knowledge creation" which e-Science makes possible: that of "comparative research" and "computational discovery". Comparative research comes about as a result of having vastly improved access to datasets, via provisions of e-Infrastructure such as digital networks and authentication mechanisms, and using computing techniques to be able to execute potentially very large numbers of comparisons between items in those datasets. In this way, new knowledge (often in the form of statistical trends) is created from the data. Similarly, some techniques which grew out of artificial intelligence research from the last fifty years (especially knowledge bases and inference systems) allow new knowledge to be inferred from e-Science's datasets using computers.

Important steps are being taken in using computers to carry out comparative research and even computational discovery (by modelling human music cognition, for example) against musical datasets. While initial results are promising, some important problems remain. Mainly they are concerned with engaging the rest of the discipline in such practices. Currently, no musicologically grounded research programme which uses e-Science and comparative techniques has been established; no major research questions have been identified.

Purcell Plus aims to address these issues in two ways. First, it attempts to produce a proof-of-concept software framework which gives musicologists a handle on musical information in forms which they find familiar (score, performance, text), but also access to computational tools for manipulating that information. Second, it attempts to engage scholars from the musicological community to discover what new research programmes may be made possible by using the techniques presented in the framework.

In this paper we present the framework itself. The project is using an example corpus of Henry Purcell's Fantazias and In Nomines, a collection of 16 works for viols composed around the 1670s. We have amassed numerous digital representations of these works in the form of encoded scores and performances, as well as numerous textual commentaries such as analyses and other scholarly literature, concert programme notes, record reviews, and sleeve notes. The framework provides an interface which gives the user access to these representations of the works in an orthogonal manner; editions can be viewed along with manuscripts, notation can be aligned with performances, and portions of texts which describe a particular point in a work can be consulted at the same time as viewing the score. The framework makes use of numerous technologies designed for dealing with music as information, especially the AudioDB musical features database, and the AMuSE framework for defining and executing queries and algorithms against musical information agnostic of any particular encoding. We describe some of the techniques we have used to develop this framework including: the RDF-based mechanisms we employ to establish linkages between representations of works (both musical and literary) and between parts of works; the querying of audio and notation; and the automatic temporal alignment of score and performance information.

A whole subdiscipline is beginning to form on the fringes of musicology. However, the practitioners of this subdiscipline are rarely working in university music departments, rather they are working in computer science and electrical engineering departments. The work they are doing is often not recognisable as musicology, but rather often looks more mathematical or more like software development. It's a culture which relishes the large scale (as opposed to musicology's close reading) and finds value in the commonplace (as opposed to musicology's interest in the exceptions). It's also a culture, however, which seeks acceptance amongst its elder siblings in the university music departments (historical musicology, music theory, ethnomusicology, practice-led research); its exponents are always keen to promote their methods and tools amongst their musicological peers.

This intensely empirical subdiscipline (computational musicology, computer-assisted musicology, music informatics), however, is facing an crucial stumbling block. An empirical practice requires evidence. In the case of the kind of music research this subdiscipline deals with, that evidence must be derived from musical information, and that musical information must be available in forms suitable for computer processing. So the problem arises at the point of attempting to collect and encode musical information.

In this paper, then, I identify and address some of the problems facing those who wish to accumulate data sets for computer-assisted music research. Such problems include: inefficiencies of data capture (optical music recognition is still a developing technology, automated transcription is impossible except for simple monophony); infrastructural difficulties in carrying out the work of building and maintaining a data set; questions of the cultural validity of a collection of musical fragments (especially the assumption of completeness of any of those fragments, their fixing in a particular encoding, assuming the power to permit their inclusion).

I examine examples of both closed and open data sets and show how these two different classes suffer from or overcome the problems I have outlined.

Purcell Plus is an eScience project funded jointly by AHRC, EPSRC and JISC which aims to link techniques for dealing with large scale datasets with traditional methodologies for musicology. Our initial dataset comprises some 300 digital objects which all relate in some way to our target corpus: Henry Purcell's Fantazias and In Nomines, a collection of 16 works for viols written around the 1670s. These digital objects include audio, music notation (either as images, or encoded using a variety of text and markup based formats), MIDI, and text and are representations not only of the works themselves, but also of literary works which describe the music and its historical context (e.g. music analytic essays, concert programme notes, record sleeve notes, etc.)

An importantly distinctive feature of our project is the integration of textual sources into a music dataset. Although considerable prior work exists in dealing with music datasets, such work tends to deal with audio and/or notation data, but not with written discourse about music.

In this paper we report on our work in integrating music and text. Of the several tasks that will enable this integration, the first we describe is that of defining formal semantics for concepts used in music theory, analysis, and historiography. Current developments in the Semantic Web include techniques for specifying ontologies, machine readable descriptions of specialist knowledge domains. We shall be building upon such projects as the Music Ontology to develop formal semantics for describing the musical concepts found in music literature.

We discuss the problem of identifying occurrences of terms outlined in our ontology in real instances of music literature, at first as represented within our dataset, but eventually with widening scope. We describe how we hope to adapt TEI markup for this task, as well as the related problem of how to link such occurrences to the musical data itself.

We have encoded the Fantazias and In Nomines in an XML format which allows us to use XPath to make references to both atomic details (such as notes) and to ranges in the musical data. Because XML is the common substrate we employ for encoding both the literature and the music, asserting references between the two does not require the implementation of any further technical infrastructure, parsers for XPath are readily available.

The software infrastructure developed for use in the Intelligent Sound and Music Systems lab at Goldsmiths, AMuSE, formalises many of the music referencing concepts which we take advantage of in this work. By providing an abstraction barrier in front of the details of how any particular encoding of music works, it allows us to implement algorithms to reason about music.

Given these semantic bases for dealing with text and music, we also discuss preliminary work in automating the referencing process through the integration of text mining (of music literature) and music information retrieval using frameworks such as SEASR. It is hoped that this work may lead on to providing an infrastructure to allow us to engage interactively with the music-scholarly community (through social networking-like techniques) and build up an evidential basis for the definition of the traditional methodologies for musicology which are the primary focus of our investigation.

We hope that this foundation may allow us to go on to provide the beginnings of an empirical basis for asking questions such as, what are musicologists able to say about music, and how might this be supported evidentially or verified?