Guide to Digital Music Editing

musica 2 Huma-Num Guide to Digital Music Editing Project directed by Philippe Vendrix (CNRS) and Augustin Braud (CNRS) $ translated by Zoe Saunders in collaboration with : $ Ailin Arjmand (University of Tours) $ Christelle Chaillou (CNRS) $ Kévin Roger (University of Lorraine) $ Suzy Piat (University of Tours) $ Vincent Besson (CNRS) $ <TOURS 2025> $ How to quote: Ph. Vendrix and A. Braud (ed.), Guide to Digital Music Editing, Tours, 2025, transl. by Zoe Saunders, https://doi.org/10.5281/zenodo.17698418 CESR-RicercarLab ISBN: 979-10-981515-1-4 2 table of contents 1. Introduction 1.1 What is Included in this Guide ? What is it not ? Who is its Intended Audience ? $ 4 1.1.1. A Resource for Design and Implementation $ 5 1.1.2. Quality Principles and the Need for Best Practices $ 6 1.2. Design conditions $ 7 1.2.1. On Scientific, Political, and Editorial Context $ 7 1.2.2. Open Access, the Cost of Print Publishing, the Market $ 9 1.2.3 Why a European Approach? The Specificity of the Field of Music $ 9 2. Music Editing and the Digital Revolution 2.1. Notational Tools $ 11 2.1.1. Operating Principles $ 11 2.1.2. Main Software $ 13 2.2. Music Encoding $ 16 2.2.1. Formats $ 16 2.2.1.1. MusicXML $ 16 2.2.1.2. MEI $ 17 2.2.1.3. HumDrum $ 19 2.2.1.4. PDF $ 19 2.2.2. Metadata $ 20 2.3. Tools for Editing and File Conversion $ 22 2.3.1 The Viewer : Verovio $ 22 2.3.2 MEIFriend $ 22 2.3.3 OCR/OMR $ 22 2.3.4. Converters: Meico and MEI Garage $ 23 3 table of contents 2.4. Copyright and Legal Concerns $ 24 2.4.1. The Role and Visibility of the Editor ; Collaborative Processes $ 24 2.4.2. Musical Editions, Digital Editions, and Rights $ 25 2.4.3. Recorded Music and the CCI $ 26 3. Design 3.1. Musical Data $ 27 3.1.1. Notational Diversity : Pedagogical and Historical Tools $ 27 3.1.2. Essential Signs versus Complementary Signs $ 28 3.2. Creating a Presentation Model $ 29 3.2.1. Annotation Spaces and Multimodality $ 30 3.2.2. Musical Analysis and Postprocessing Encoded Files $ 31 3.3. Relationships with Sources $ 32 3.4. Data Lifecycle, Management, Accessibility, and Reuse $ 34 3.4.1. Digital Edition or Database $ 35 3.4.2. Data Sustainability and Visibility $ 36 4. Conclusion 4.1. Checklist for Digital Editions; What are the Essential Steps ? $ 37 4 1. introduction Digital editions of music occupy an increasingly prominent place in the international musicological landscape. While it is impossible, due to the lack of exhaustive catalogues,1 to enumerate them, their presence is becoming more significant. Large-scale national and European projects dedicate substantial portions of their programs to the production of digital editions. This enthusiasm, though still far from comparable to the digitization of recorded music or musical sources, raises questions for both their creators and users.2 This guide aims to address some of those questions. Developed from reflections initiated in 2023 within the Consortium-HN MUSICA2 of the research infrastructure HumaNum,3 it is intended to be regularly updated to incorporate innovations that affect the world of music and digital technology, and aims to propose a framework to counter obsolescence by recommending workflows that ensure the longterm sustainability of data and the infrastructures that host it. The guide aims to accompany the essential stages of a digital music edition, from its conception to its dissemination. It therefore includes information, recom1 The contribution of the Directory of Digital Scholarship in Music, compiled by the Digital Humanities Interest Group (DHIG) between 2018 and 2023, should be mentioned : https://rutgersdh.github.io/musicdh/reuse/ 2 Two research seminars were organized in Tours, from July 3-5, 2023, and February 11-12, 2025. Participants included Nicolas Bucher, Laurent Guillo, Theodora Psychoyou, Thomas Lecomte, Saraswathi Shukla, Christelle Chaillou, Christelle Cazaux, David Chappuis, Rémi Bonnin, Richard Freedman, Léontine Fortin, Carlo Bossi, Marco Gurrieri, Kévin Roger, Laurent Pugin, Ugo Bindini, Aurélien Balland-Chatignon, Philippe Vendrix, Augustin Braud, Ailin Arjmand, and Vincent Besson. 3 See https://www.huma-num.fr/quest-ce-que-l-ir-huma-num for more information on this research infrastructure. 4 A mode of disseminating research which results in a free digital format, in compliance with copyright law. 5 For more information on the Cultural and Creative Industries, see the program pepr-iccare : https://pepr-iccare.fr/about-iccare/project/?lang=en. mendations, suggestions, and guidance addressing both the definition of the corpus to be edited and the choices involved in web-based visualization. It attempts to account for the evolving context in which musicology operates, both in terms of its ties to musical practice and its positioning within the cultural production chain. In a context in which an increasing number of projects require a strong digital foundation in line with Open Access4 policies and the expectations of research funding bodies, the following pages are intended as a resource for identifying the scientific actions necessary to meet current challenges and to emphasize the critical role of the underlying logistics that support the technical and scholarly aspects of such projects. Although musicology is not itself a Cultural and Creative Industry (CCI), it is closely related to one of that sector’s most dynamic and diverse fields, audio recording and distribution. In other words, while this guide primarily concerns heritage science, it also addresses the role that musicology as a scientific discipline seeks to play within the rapidly expanding realm of CCIs.5 1.1 What is Included in this Guide ? What is it not ? Who is its Intended Audience ? 5 1. introduction Heritage sciences, production chain. These terms alone convey the scope of the field covered in this guide. It is therefore necessary to consider certain notions and concepts before beginning any editorial project. Nothing in this domain is unequivocal : the fact that a musical work is considered part of our heritage does not mean that it is free from legal constraints ! A production chain may choose not to use subscription-based tools. The act of editing, digital or not, takes place within a social, economic, legal, and cultural context that invariably requires preliminary reflection. Under European law, the status of a scholarly editor is distinct from that of a creator, but it is precisely this distance that must be assessed with care before undertaking an editorial project, as it can easily give rise to difficult and potentially damaging conflicts for the scholarly project.1 At the conclusion of the project, the question of obsolescence will inevitably arise. Like legal considerations, it lies beyond the scope of this guide, yet it remains a fundamental, albeit intangible, element upon which the entire project must be constructed.2 Any editorial project requires both human and financial resources. It remains difficult even today to quantify in hours or material costs the production of a digital musical edition.3 An estimate based on the Gesualdo Online project values a single page of music (in A4 format) at approximately 250 euros, which the entire process from the acquisition of source materials to the online publication of files in a functional and userfriendly format.4 Undertaking a digital edition therefore 1 On the question of music publishing rights, see Allen Bargfrede and Cecily Mak, Music Law in the Digital Age, Boston, Berklee University Press, 2009. 2 On obsolescence, see Rosenthal, D.S.H., “Format Obsolescence: Assessing the Threat and the Defenses”, Library Hi Tech, 28/2 (2020), p. 195-210. 3 For a European approach to the subject, see European Commission, Directorate-General for Research and Innovation, Kramer, B., Study on scientific publishing in Europe – Development, diversity, and transparency of costs, Publications Office of the European Union, 2024, https://data.europa.eu/doi/10.2777/89349. 4 Ph. Vendrix, Gesualdo online, the economic model of a digital music edition, 2024, http://doi.org/10.5281/zenodo.13959918. requires mobilizing resources and, if necessary, obtaining funding from sponsors. This guide will not provide a fixed “grid” of such costs, which would inevitably vary depending on the type of corpus, from a plainchant antiphon to a 21stcentury opera, from a popular song of the second half of the 20th century to a 17thcentury collection of organ tablatures. Assessing the corpus is a crucial step that involves considering the feasibility of the project, whether it requires the participation of numerous specialists or to can be completed by a single researcher familiar with the repertoire in question. From these considerations, a financial appraisal of the editorial project can be drawn, which would be included in any funding application, whether the project is purely editorial or part of a broader scientific project. 1.1.1. A Resource for Design and Implementation Although conceived as a resource to support the design of digital editions, this guide does not claim to address the technical issues that inevitably arise and recur with each editorial project. It is not limited to meeting the requirements of paleographic work and or collating sources ; it also anticipates the ways in which the text to be transmitted will be used through its reading formats, the role of paratext and metadata, and the management of notation systems employed in both early and contemporary music. Accordingly, this guide will not address specific cases, but rather outline the full range of issues that should be considered when undertaking a digital 6 1. introduction music edition. It is therefore structured around three main themes: (1) the nature of a digital edition; (2) the tools and processes required for its production; and (3) recommendations for an efficient design that meets the FAIR1 principles that govern any digital humanities initiative. Addressing these three themes raises a series of questions of varying level of detail, but whose implications are significant for the nature of the results obtained through what is often a lengthy process. This guide provides responses to both general and specific questions, including: $ How is music encoded? $ Are there legal constraints that apply to a digital edition? $ How can critical thinking be applied to a digital edition? $ How can the digital twin represented by the new edition be linked to its source and its notation? $ What connections should be established between a digital edition and other available digital resources (RISM, Gallica, etc.)? $ How can the planned editorial work be aligned with European cultural policies? $ On what grounds can musicology be considered a data science? $ What relationships can a digital edition forge with the CCIs, and with music platforms in particular? $ Does a community of music editors exist? In order to respect best practices, it is necessary to establish recommendations for monitoring the quality of both data and encoding. This guide therefore encourages 1 The FAIR principles are a group of recommendations that aim to improve accessibility, interoperability, and reuse of digital resources. The original publication of these principles was drawn up in 2016 in the journal Scientific Data, see M. Wilkinson, M. Dumontier, I. Aalbersbert et al. “The FAIR Guiding Principles for scientific data management and stewardship.”, Sci Data 3, 160018 (2016). https://doi.org/10.1038/sdata.2016.18. researchers to exercise critical judgement and to assume full editorial responsibility in order to ensure that Open Access policies are upheld through scientific expertise. One of the fundamental principles of digital editing is to preserve the integrity of the content while providing it a renewed perspective. From a philological standpoint, digital editions raise a number of questions, especially concerning critical editions. When preparing a digital critical edition, it is essential to consider the computational readability of the data, and therefore its encoding, as well as usability for performers. Since these two dimensions often pull digital editorial work in different directions, they should be approached in a complementary manner to enhance intelligibility, clarify objectives, and anticipate the contexts in which the edition will ultimately be used, whether for research, performance, etc. Because digital editing is far more complex than simply reproducing musical examples or excerpts intended for distribution as downloadable .pdf files, particular attention must be paid to achieving a high level of interactivity through features specifically adapted to encoded musical text. Thus in the course of developing a digital edition, scholars must ensure compliance with the FAIR principles, while also assessing the technical feasibility of publishing the data online and maintaining its accessibility over time. 1.1.2. Quality Principles and the Need for Best Practices The guide is intended for musicologists and musicians working in teaching and research institutions as well as for any knowledgeable individual, professional or 7 1. introduction amateur, wishing to venture into digital editing. In Europe, the former are generally encouraged to disseminate their work in accordance with Open Access principles, whereas the latter typically operate independently and can be broadly divided into two categories : the first makes their editions freely accessible through dedicated websites, while the second produces digital editions for strictly limited use (within a performing ensemble, for example). Evaluations of Open Access projects often focus on what can be described as the quality process. $ How is the peer review, which is essential to any scientific process, conducted? $ How can one balance making the resource fully accessible and preserving the expertise involved? Attempting to answer these questions requires situating the digital edition within the musical philological tradition of the second half of the 19th century. It is also important to examine the relationship between a digital edition and what is conventionally referred to as a critical edition, since the critical edition is by no means the only point of reference for those who produce digital editions. Establishing connections with editorial traditions is especially pertinent because, as this guide illustrates, the three pillars on which a digital edition rests—tools, content, and presentation1—are conceptually closely aligned to what the founders in the 19thcentury conceived. This conceptual proximity calls for careful nuance : $ How does downloadable .pdf file differ from a printed edition? $ How can it be characterized digitally? $ Are there multiple levels of digitalization for a musical edition? It thus appears necessary to provide a guide to clarify what “digital” truly means: $ Are “digital” editions intended to be understood by computers or by performers? $ What is the defining feature of a digital edition in terms of intelligibility? 1 See M. Caraci Vela, Musical philology. Institutions, history and critical approaches, Pisa, Edizioni ETS, 2015. We wish to emphasize a particular point of caution regarding the rendering of encoded files. While the prospect of an aesthetically pleasing visualization is certainly appealing, it may result in encoding that is less readable and less logical for machines, thereby limiting the file’s interoperability and the potential for performing analytical tasks that make full use of the digital edition. One best practice would therefore be to encourage users to prioritize the use of established standards in order to ensure maximum readability and interoperability of files. 1.2. Design conditions If we consider only the field of digital editing produced within the framework of institutional research (carried out by members of recognized teaching and research institutions as defined by European criteria), it is essential to address the conditions under which editorial projects are conceived, often in response to specific requirements set out in calls for proposals. Beyond these particular cases, shaped by the nature of the academic or research institution, as well as by local conditions of scientific production and dissemination, it is equally important to situate the dynamics of digital music editing within a broader perspective that connects European research with the wider field of music scholarship. 1.2.1. On Scientific, Political, and Editorial Context Scientific knowledge today is built largely upon digital resources ; this is as true for disciplines such as economics or archaeology as it is for musicology. Some of the most emblematic projects initiated by the musicological community in the mid20th 8 1. introduction century—namely RISM and RILM—have long since undergone their digital transformation. It would now seem unusual to conduct original research without relying on online resources in one way or another, just as disseminating research through journals or publishers can no longer dispense with digital tools. Visibility of research results is the primary and perhaps most compelling reason for this shift, though not the only one. A recent study conducted by COST EarlyMuse on the state of monumental editions in Europe has in fact identified numerous arguments in favor of digital editions, even when they concern “historical” projects.1 Beyond the large-scale “monumental” projects, some of which have been ongoing since the 19th century, the question of the economic context has at times become particularly pressing. Rising paper costs, reduced library acquisitions, distribution expenses, and shifting market trends all combine to make the work of a publisher extremely challenging and risky when it comes to producing music volumes. The distribution of forces and dynamics in the publishing sector are similar to those of the music recording industry. Production and distribution channels increasingly converge under the control of a few major players capable of sustaining full production costs. Yet even these powerful enterprises cannot disregard the contribution of scholars, and thus the value of their research, which is generally supported by public institutions. It would be utopian to imagine complete independence among the various parties involved : the scholar who prepares the 1 A. Puentes-Blanco, M. Kokole, Ph. Vendrix, M. Gembero-Ustárroz, R. Herissone, C. Troelsgård, K. Grabnar, B. Lodes, K. Schöning, & V. Vrbanic (2024). The monumental edition in the digital age: creating a sustainable future. https://doi.org/10.1080/09298215.2024.2373998. 2 On this topic, , see G. Born, “For a Relational Musicology: Music and Interdisciplinarity, Beyond the Practice Turn: The 2007 Dent Medal Address,” Journal of the Royal Musical Association, 2010, Vol. 135, No. 2, pp. 205–243. 3 For a detailed discussion of this point, see Ph. Vendrix, A. Puentes-Blanco, V. Borghetti, A. Piéjus, D. Lewis, A. Olmos, N. Marjanović, Ch. Riedo, G. Pranis, V.Parilla, & V. Vrbanic (2024). Musicological Publications in the Digital Age: Charting a New Course. Zenodo. https://doi.org/10.5281/zenodo.13961016. edition, the policymaker who facilitates it through institutional support, and the publisher who ensures its production and dissemination. To this trio must now be added a fourth and essential actor: technology. The context with all its complexity invites a rethinking of methods, objectives, as well as of content and its modes of presentation. Such reflection, however, should not be limited to those producing digital critical editions ; it must also include users, encouraging them to create a new relationshp between musicology and performance. In other words, it is equally necessary to consider the visibility of musicology as a scientific discipline and its impact on the world of musical performance.2 In this respect, it is worth noting the significant scale of editorial projects in Europe as well as their importance in terms of budgets allocated to them by local, national, and European public authorities.3 Printed and digital editions alike incur production and maintenance costs. Yet it would be overly simplistic to reduce the distinction between these two modes of disseminating knowledge to this sole opposition. The impact of digital technology on the cultural and national heritage represented by music can provoke genuine ideological debates, as illustrated by the example of the monumental Chopin editions, where two different versions were published. Policies and marketing are closely intertwined, with increasing offerings and funding opportunities for editions. Although monumental music editions are still of great importance, 9 1. introduction they are difficult to justify because they are costly for publishers – through salaries, materials, etc. – as well as for potential buyers. Hence the need to reassess editorial methods and to redefine the objectives of new editions. From the early days of Open Access to more recent developments, the methods for rethinking the economic model and the creation of new modes of dissemination have evolved considerably. Numerous avenues nevertheless remain to be explored. Yet a crucial question persists : how can we convey the importance of these new editorial practices to concert organizers, performers, indeed the broader musical community, as well as to the world of publishing and libraries? 1.2.2. Open Access, the Cost of Print Publishing, the Market Music editing has been reinventing itself in recent years through ambitious scholarly projects devoted to the digitization of sources. Beyond the obvious reduction in printing costs, digital editions offer unprecedented longevity and accessibility to centuries old sources, while also creating new opportunities for computational analysis. European recommendations increasingly favor sharing resources in Open Access : by adhering to these standards, research results can be made readily available to the scholarly and musical communities in a more sustainable and economical way. By offering a heritage science response to the challenges posed by the digital humanities, digital editing rethinks the very notion of authority by fostering dialogue, whether through new reconstructions of missing parts or through the exploration of the various versions of a single work. The possibility of viewing of the critical apparatus, multiple versions, links to metadata, and audio recordings 1 R. E. Scott and A. Shelley “Music Scholars and Open Access Publishing” (2022). Faculty and Staff Publications – Milner Library. 161. https://ir.library.illinoisstate. edu/fpml/161 on a single web page drastically reduces editorial costs compared with those of printed editions, while the flexibility of the digital medium facilitates collaborative work. It is also possible to modify the layout of the score dynamically and to conduct analytical research within XML files using scripts. In addition to statistical analyses, that nonencoded formats cannot support, even simple variations in layout are impossible to implement on a case-by-case basis in print editions, which must adhere to standardized conventions. The possibility of editorial “customization” tailored to individual needs is therefore more economical (financially and in terms of time) than the manual modifications or pasted additions familiar to many performers. Musicians often prepare their own page layouts when working on pieces where, for example, page turns are impractical : digital media thus offer the considerable advantage of being adaptable to the needs of musicians, particularly when using tablets. These new, lowcost editions increase the potential for disseminating and reusing data, thereby contributing to the construction of a shared foundation for scientific knowledge. This situation illustrates the persistent divide between publishers and musicians, a gap that digital editing seeks to bridge, though it remains important to keep in mind that the reception of such practices varies widely.1 1.2.3 Why a European Approach? The Specificity of the Field of Music Digital critical editions play a crucial role in the preservation of musical heritage and are often funded through national or European calls for proposals. Consequently, the question of their accessibility is of great importance. While Open Access poli- 16 2. music editing and the digital revolution Noteflight: Unlike the softwares described above, Noteflight is accessible directly through a web browser, thus does not requiring installation (fig. 6). Although it supports export to MusicXML, it remains comparatively limited relative to its competitors.1 Programs designed for viewing and editing tablature for lute and related historical instruments include LuteScribe, TabCode, and Fronimo, as well as Gregorio for neumatic notation. For XXIe music, composers increasingly work directly with image-editing software such as Illustrator or Photoshop, an approach that diverges significantly from processing encoded data and metadata. Furthermore, despite the diversity of these programs, some open source, others proprietary, all of them feature export systems robust enough to enable the exchange of musical content without substantially altering the files. These tools make it possible both to produce highquality data encoding and to share data in accordance with FAIR 1 https://www.noteflight.com/ 2 To understand the significance of linked open data and the possibilities it offers for connecting corpora, see : https://lod-cloud.net/clouds/lod-cloud.svg principles. They allow users to capture musical data by integrating personalized parameters while remaining open to other possibilities through Linked Open Data, a method designed to facilitate the sharing of interconnected data.2 2.2. Music Encoding 2.2.1. Formats Each program, whether proprietary or not, generates its own file type: Finale creates .mus files, Sibelius .sib files, … etc. In order to facilitate sharing essential data with the community, developers have consistently sought to provide export formats that enable file sharing among users working in different environments. Most of these programs now offer several export options, including MusicXML, pdf, midi, image formats, and MEI either directly within the application or through plugins. 2.2.1.1. MusicXML Although pdf format remains widely used for commercial publishing and printing, open export formats such as MusicXML enable computational analysis and ensure data interoperability. They are therefore of clear interest to both musicologists and publishers, who can use them to share files easily. Based on the universal XML textual format for transmitting and disseminating data, MusicXML is a structured, hierarchical, and extensible language that nevertheless has one drawback – it is verbose : one measure may correspond to several hundred lines of code. The vast majority of the software discussed above supports native export in MusicXML, making it a widely accessible and high-quality standard. However fig. 6: overview of Noteflight workspace 17 2. music editing and the digital revolution some data and metadata can be omitted during export to MusicXML format, which necessitates additional work on the resulting files. It is therefore essential to pay close attention to the original encoding upon which the data export processes depend. Good encoding practices greatly facilitate the overall editing process by avoiding unnecessary accumulation of elements that could compromise the encoding, for example, an excessive use of personalized or unconventional graphics. 2.2.1.2. MEI MEI (Music Encoding Initiative) is an encoding format designed to support the semantic and scholarly use of musical data. It builds on MusicXML format and is directly derived from TEI (Text Encoding Initiative), one of the earliest specialized text formats, created in 1987 for the critical and diplomatic editing of texts. MEI was introduced in 1999–2000 as an open source format for music encoding that can generate PDF and SVG files (viewable through online rendering tools such as VexFlow and later Verovio), as well as MIDI files. Similar to TEI, a single master file can now serve for both online publication (webXML) and a printed score. The strength of encoding, and the MEI format in particular, for music editing lies in the coexistence of two complementary dimensions : a graphical output for visualization of the score, particularly useful when working from manuscripts, and a logical transcription of musical information into code, which can thus be subjected to analytical processes and networked parametrically (fig. 7). These elements are organized logically and structured clearly, making MEI more suitable for processing musical data than the XML format. Indeed, XML requires a separate line for each descriptive element, whereas MEI consolidates all this information within a single line. This economy of notation translates into an economy of storage, thanks to a more compact format (fig. 8): fig. 8: the same passage, shown in XML format on the left and MEI format on the right. fig. 7: a Renaissance polyphony encoded in MEI and its graphic output, both visible within a single interface. 18 2. music editing and the digital revolution fig. 9: Avava Inouva, Idi (modern Kabyle song) shown alongside ; exported in .mei format with the text integrated.1 fig. 10: FRBR conceptual model.2 1 See S. Martin de Guise. La MEI (Music Encoding Initiative) : un standard au service de la musique kabyle. Iles d’Imesli, 2013. hal-02948837 2 B. B. Tillett (Library of Congress) : “What is FRBR ? A Conceptual Model for the Bibliographic Universe,” published in Technicalities, vol. 25, no 5 (September 2003). 3 https://github.com/music-encoding/sibmei 4 For more information, see : https://music-encoding.org/guidelines/v3/content/frbr.html Several tools are available to export files encoded in MEI. Although Sibelius is no longer actively developed, it remains widely used in the editorial community. Exporting to MEI is straightforward via the Sib>MEI plugin developed by the MEI community and available online,3 along with the necessary installation instructions. However, it appears that the preferred software for digital music editing is now MuseScore, as it allows direct export to MEI from the application without the need for a plugin. Data preservation is generally satisfactory for traditionally notated repertoires, and it can also accommodate oral music traditions by enabling clear visualization of textual elements (fig. 9, opposite). MEI format also allows for the inclusion of a critical apparatus, potential variants, and an extensive set of metadata within its header, which can, for example, integrate the FRBR model (Functional Requirements for Bibliographic Records) (fig.10) to establish more precise connections with the library world.4 One of the most important features of MEI format is its ability to designate noteworthy elements using the attribute <xml:id>, which allows each element to be uniquely identified. These elements can then be linked to sources and audio recordings through thesauruses and reference systems. For certain notational symbols, such as those used for musica ficta, it is even possible to automate some necessary interventions in the code to ensure a standardized visualization. A short Txi lek - el li - yin ta burt - a Va vaI - nou - va - a Va vaI - 9 10 11 &#Em % œ‰œjœ œ œœ œjœœjœœœœjœœjœœœ 19 2. music editing and the digital revolution Python1 script can be used to enrich or correct files, provided they have been properly prepared in advance. Numerous editorial projects already rely on this format.2 Because it is user friendly, has an active community, and can illuminate both musical heritage and its associated data, MEI has become the preferred solution for digital music editing. 2.2.1.3. HumDrum Humdrum is a suite of tools and resources for computational music analysis, created by David Huron in 1995 and maintained by Craig Sapp at the Center for Computer Assisted Research in the Humanities. Its syntax provides a data structure within which users define specific interpretive schemes with a level of complexity appropriate to a given corpus. The **kern music notation format is used regularly in various projects3 ; it allows for the representation of basic elements inherent to Western music, such as pitch and duration. However unlike many other music formats, “humdrum is not limited to traditional Western notation, but rather affords the study of nonWestern, vernacular, or avantgarde musics, as any type of musical feature, data, or metadata to be easily encoded in the same place.”4 This flexibility makes it possible to develop custom tools to address specific musicolog1 Python is a general-purpose programming language that is suitable not only for software or web development but also for computational operations on corpora, as is the case here. 2 For example, Gesualdo Online and Corpus Monodicum. 3 One recent example is the encoding of the first Chopin editions, C. S. Sapp & M. Konik (2025). pl-wnifc/humdrum-chopin-first-editions: Humdrum digital scores of Chopin first editions (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.14879352 4 N. Condit-Schultz & C. Arthur. (2019). humdrumR: a New Take on an Old Approach to Computational Musicology. Proceedings of the 20th International Society for Music Information Retrieval Conference, 715-722. https://doi.org/10.5281/zenodo.3527910 5 https://www.humdrum.org/ ical questions by encoding directly in MEI and using a compatible viewer, such as the Verovio Humdrum Viewer (fig. 11). The Humdrum encoding format is ideal for the analysis of large datasets due to the flexibility of the tools it offers. fig. 11: Die Kunst der Fuge, Contrapunctus 1, a 4., J-S Bach ; humdrum visualization using the Verovio Humdrum Viewer.5 2.2.1.4. PDF Although formats such as XML or MEI enhance analytical possibilities, many users predominately use PDF. The pdf format is central to the work of commercial 20 2. music editing and the digital revolution publishers as well as performing musicians, who value it for its ease of use and the ability to make handwritten annotations on tablets. Companies such as Newzik1 have also recognized the value of a tool that for annotating and organizing scores, including in ensemble work. It is important to note, however, that the mere conversion of a manuscript into a .pdf file does not constitute a true digital edition, since it does not allow one to take advantage of the intrinsic properties encoded in the file, despite its satisfactory graphical output. 2.2.2. Metadata One of the main challenges of digital music editing is creating a functional and enduring link between metadata and musical data. Various encoding processes make it possible to add a wide range of metadata and to attach them to targeted elements : MEI (for music encoding) and TEI and LaTeX (for text) are particularly well suited to these purposes. To consider the essential metadata for ensuring the quality of an encoded file, it is first necessary to refer to the MEI Guidelines.2 Metadata are gathered in the header,3 which provides a comprehensive description of the digital file and must a minima allow for the creation of an appropriate bibliographic citation. This typically includes the file title, information about the source from which the electronic document was encoded, and details regarding the MEI elements used. The header is structured into three major sections within a <fileDesc> : <titleStmt>, <resp1 https://newzik.com/fr/ 2 We refer to the following page within this section : https://music-encoding.org/guidelines/v4/content/metadata.html 3 Term referring to the file header in which all metadata are declared, distinct from the “body,” which contains all the descriptive elements of the music. 4 https://demo.beethovens-werkstatt.de/index.html stmt>, and <sourceDesc>. <titleStmt> contains information on the title, subtitle, and composer. <respStmt> identifies the editors, and <sourceDesc> holds information related to the source. A recommended best practice is to consult established reference frameworks when completing the header to maximize the interoperability of the edition. MEI headers that contain URIs can then be directly linked to the manuscript via Verovio, as is the case with Beethovens Werkstatt (fig. 12). fig. 12 : Autograph op. 111., Ludwig van Beethoven ; visualization of the source and the MEI header.4 21 2. music editing and the digital revolution It is, however, important to consider the purpose of the edition when deciding whether to integrate metadata directly within the encoded files, or whether their volume and complexity necessitate creating a relational database. The advantage of MEI lies in its ability to consolidate a large amount of information within a single file that provides access to metadata and a highquality graphical representation. Some research projects will indeed favor files rich in metadata through links established with reference systems (VIAF, Wikidata, Geonames, Getty AAT, and MIMO, for instruments), while others may prefer a more streamlined approach to facilitate seamless analysis across large corpora. In the case of music, it is a recommended best practice to register metadata with RISM, regardless of historical period. A RISM ID1 allows for the linking of key information, including the type of source, the individuals who edited or annotated it, and the holding institution (fig. 13). 1 For more information on this topic, see https://rism.info/community/how-to-cite-rism.html fig. 13 : De double dueil, Jean Servin ; export in .mei format, header incorporating metadata connected to established reference systems. 22 2. music editing and the digital revolution It is also possible to rely on the FRBR model, a conceptual framework for bibliographic data that organizes works and sources. This model can be integrated into the header1 to facilitate communication with libraries – which constitute the majority of corpora – and their standards. Additionally, use of the Semantic Web enables edited music to be incorporated into European projects such as the forthcoming European Cloud for Cultural Heritage, which is based on the CIDOCCRM2 ontology and its LRMoo extension (itself grounded in the FRBR model).3 Although this constitutes a relatively late stage in a digital editing project, it is important to consider the interoperability of the corpus from the outset to ensure its effective dissemination. This objective is crucial for the visibility of our digital music editions as well as for musicology. 2.3. Tools for Editing and File Conversion 2.3.1 The Viewer : Verovio Verovio is an online tool that displays MEI files automatically in score format, using SVG (Scalable Vector Graphics). This viewer is developed by the RISM Digital Center with the support of the Swiss National Science Foundation. Verovio is written in standard C++ and can be compiled either as a standalone commandline tool or as JavaScript. Beyond simple score visualization, its main advantage is 1 On this topic, see https://music-encoding.org/guidelines/v4/content/metadata.html#FRBR 2 For more information, see : https://cidoc-crm.org/ 3 For more information, see T. Bottini, A. Braud, M. Gurrieri et K. Roger : Approches de la complétion des Headers MEI par les GT1>2 du Consortium-HN Musica2: https://github.com/Amleth/consortium-musica2-gt2-ontologies/blob/main/modules/livrables/mei_headers_gt1_gt2/index.md 4 On mei-friend, see : https://mei-friend.mdw.ac.at its accuracy in preserving the metadata of the original MusicXML file, as well as its ability to edit and correct MEI files in real time. Despite its user-friendly design and the immediacy of visual rendering, thorough preparation of files prior to MEI export remains essential to ensure high-quality output. Verovio do not yet support all aspects of advanced notation, in particular the graphical features required for the transcription of medieval repertories and the symbols used in contemporary music. Particular care must also be taken with works containing text, ensuring the precise alignment of syllables with their corresponding notes. 2.3.2 MEIFriend MEIFriend is an editor for musical encodings hosted at the University of Music and Performing Arts Vienna, in Austria. It is a browser-based application that uses Verovio as a viewer. It enables users to edit and validate MEI files directly within the web browser. This platform is particularly well suited for realtime visualization of any modifications made to the MEI code.4 2.3.3 OCR/OMR One of the most significant recent innovations in music editing is the development of OCR (Optical Character Recognition) and OMR (Optical Music Recognition) technologies, which are increasingly supported by artificial intelligence. 23 2. music editing and the digital revolution The U.I. for Computing Research at the University of Alicante in Spain1 has made substantial developments in OMR, while in France, Collabscore was one of the first projects to employ this technology. Its OMR system combines deep learning with tools for modeling music notation: “The system analyzes the image of the score and, drawing on its knowledge of musical notation, is able to produce a complete representation of the recognized score while verifying its internal consistency.”2 The DMOS OMR is thus capable of distinguishing parts from voices, as well as visualizing the text and linking it to corresponding staves (fig. 14) : fig 14 : two examples : voice identification across multiple staves (l) and text recognition (r).3 1 C. Penarrubia, C. Garrido-Munoz, J.J. Valero-Mas, and J. Calvo-Zaragoza,“Efficient Notation Assembly in Optical Music Recognition”, in Proc. of the 24th Int. Society for Music Information Retrieval Conf., Milan, Italy, 2023. 2 https://collabscore.github.io/collabscore.html 3 For more information, see https://collabscore.github.io/omr.html 4 https://www.aruspix.net/index.html 5 For more information, see https://github.com/Audiveris/audiveris 6 A complete list of resources on OMR can be consulted via this link : https://github.com/OMR-Research/omr-research.github.io/blob/master/OMR-Related-Research.bib For early music, Aruspix,4 developed by Laurent Pugin in connection with the Marenzio critical edition and in collaboration with the MEI community, aims to maximize interoperability. Aruspix allows users to superimpose two copies of the same work, to identify the symbols used in historical compositions via an OMR algorithm based on hidden Markov models, and it supports collatio of different sources. Audiveris,5 a free, fast, and powerful OMR tool, is a more accessible application used by MuseScore. It is recommended for work on sources with traditional notation, presented clearly – for example the corpora in resources such as IMSLP – and can serve as a solid starting point for employing such tools in a digital edition.6 The use of OMR tools highlights the importance of decisions made by the scholarly editor, decisions that extend beyond philological debates to encompass broader ethical considerations, the urgency of which has increased with the proliferation of artificial intelligence and its social and environmental impact. 2.3.4. Converters: Meico and MEI Garage Encoding is useful beyond visualization and metadata referencing. Depending on the project, it may also be advantageous to export an edition in various formats, 24 2. music editing and the digital revolution such as an audio file or even a REST API (Application Programming Interface). One of the most effective tools for working with MEI files is meico, a program written entirely in Java. Although MEI has become a standard for digital music editions, processing MEI-encoded music is not a trivial task, and many application scenarios rely on more established and efficient formats. Meico implements methods to convert MEI data into other formats, making MEI encodings accessible to a wider range of applications.1 MEI Garage2 is another tool that facilitates common tasks such as converting between formats like MIDI, as well as modifying MEI files within a corpus. Its clear and user-friendly interface makes it accessible even to beginners. 2.4. Copyright and Legal Concerns 2.4.1. The Role and Visibility of the Editor ; Collaborative Processes Encoding scores allows for the inclusion of metadata, such as the roles of editor, engraver, and reviewer. The role of collaborators in music editing raises fundamental questions about the boundaries between individual contributions and collective efforts. How should one distinguish between a collaborator and an editor or commentator? For example, should someone who corrects individual notes or articulations be considered a collaborator, or does their role fall under editorial assistance? 1 https://github.com/cemfi/meico?tab=readme-ov-file 2 https://meigarage.edirom.de/ 3 Abbreviation for : Responsability Statement 4 As with, for example, Gesualdo Online. MEI format offers a partial solution to these questions through <respStmt>,3 included in the header discussed above, which allows for listing all contributors who have worked on the edition (fig. 15) : These tools are particularly important in the context of online music editions that allow for the completion of missing voices.4 This distinction carries significant implications for how contributions are identified in music editing and in musicology more broadly. Careful attention is required to ensure that each contributor’s input to a digital edition is properly credited. The use of identifier systems such as ORCID provides an additional means of acknowledging authors’ work. fig. 15 : Leve le cœur, Jean Servin ; excerpt from the MEI header, specifically <respStmt>. 25 2. music editing and the digital revolution Edited works serve as valuable pedagogical tools, providing performers and students with informed interpretations of the repertoire. It is not uncommon for editions to result from collaborations between researchers, musicians, and experienced engineers, as well as master’s or doctoral students engaged in research projects. These students may initially train in digital editing practices before applying this experience to their own musical and scholarly projects ; it is therefore essential to credit their contributions properly, since such collaboration can be beneficial for their future careers. Depositing projects under a Creative Commons license such as CC BYNCSA 4.0 provides legal protection for the editions1 : the author must always be credited (BY), use must be noncommercial (NC), and derivative editions are permitted (SA), a provision particularly well suited to critical editions. Highquality editions leave a lasting impact on the field of musicology : notable examples include the Josquin Project, one of the first digital music editions, as well as the CRIM, Duchemin, and Gesualdo projects.2 These editions also directly influence performers’ choices and the accessibility of works, thereby shaping concert programming and the broader music industry. Moreover, the role of experience in developing knowledge and understanding of repertories cannot be overstated. An editor’s or collaborator’s familiarity with a given repertoire often enables them to give works that might initially seem less accessible the attention they merit, thus broadening the horizons of performers and audiences. It is also important to clearly distinguish the roles of composer, editor, and arranger ; these considerations should be addressed from the outset of an editorial project to ensure that the processes that led to the final edition are transparent. 1 See : https://creativecommons.org/licenses/by-nc-sa/4.0/deed.fr 2 https://ricercardatalab.cesr.univ-tours.fr/fr/projects/ 2.4.2. Musical Editions, Digital Editions, and Rights Financial models for musical publishing depend on the correlation of many factors, the most fundamental of which is undoubtedly the scholarly quality of the edition. Beyond musicological expertise, the integration of contributors into networks for the dissemination of research results plays a crucial role in the production and diffusion of critical editions. Some institutions, such as the CMBV (Centre de Musique Baroque de Versailles), operate under commercial publishing agreements. Revenue generated from the sale of critical editions supports employment of a dedicated editorial team. While this model effectively sustains both scholarly excellence and cost management, it raises questions about data accessibility at a time when Open Access is increasingly regarded as the ideal standard. The CMBV addresses this issue through a hybrid model : some scores are made freely available online, while others must be rented for use in performances. It remains difficult to envision how monumental editions might break away from print format, as these editions still comprise extensive volumes encompassing the complete works of a composer, without even considering their high production costs : We can observe, though, that the creation of full digital editions within monumental series is still experimental and in progress. [...] In most of the cases considered here the move to digital editions would not yet be possible in a near future. [...] Skills for music editors in the digital era include for instance mastery of different music edition software and even programming knowledge, but there is a gap between musicological curricula in many universities around the world and the technological skill 32 3. design thematic catalogues within a specific corpus by grouping pieces according to key, meter, etc.1 CRIM Intervals, which builds on Music21, provides sophisticated identification of melodic, harmonic, and contrapuntal motifs characteristic of Renaissance polyphony. It was developed within the project Citations: The Renaissance Imitation Mass2: fig. 21: An example of pitch visualization for a work, generated using Music21.3 1 For more information, see https://www.music21.org/music21docs/about/what.html 2 See https://crimproject.org/ 3 For the complete example, see the following link : https://www.music21.org/music21docs/usersGuide/usersGuide_15_key.html 4 A concrete example of analysis can be consulted via this link : http://recherche.ircam.fr/equipes/repmus/Analyse/Humdrum/humdrum_exemple.html 5 https://www.haverford.edu/users/rfreedma 6 For more information, see https://github.com/RichardFreedman/mei_tools/blob/main/README.md 7 For more information, see https://mermeid.edirom.de/manual/basics.html As noted above, Humdrum is well suited to analysis due to its system of commands, which can be combined with one another. It is not limited to the tonal analysis of music : it can also detect patterns by systematically examining intervals. In addition, it enables similarity measures (correlation, edit distance), and performs editing operations using settheoretic formalism,4 which makes it particularly well suited to more complex corpora. Regarding postprocessing, Richard Freedman5 has also developed several tools (mei_tools6) that enable the simultaneous updating of metadata across a large number of MEI files from a single document, as well as the correction of minor engraving errors introduced by software idiosyncrasies. Another notable tool is Mermeid,7 which supports the management, (pre)visualization, and manipulation of metadata embedded in files. These tools can be particularly valuable when working with dense corpora or when mass modifications are required. 3.3. Relationships with Sources The digital lens allows us to rethink the critical edition by providing direct access to sources via the IIIF protocol. This is a standardized methodology for sharing, annotating, and preserving images, which appears particularly well suited to the digitization of musical sources. Indeed: 33 3. design IIIF allows for the efficient comparison of musical scores and notations across various collections, significantly easing the analysis of compositional similarities, stylistic evolutions, and regional variations. It also empowers musicologists to access, study, and share high-quality images of medieval music manuscripts that might otherwise be confined to their physical locations.1 Between the digital rewriting of a source and the digitization of the original manuscript, it is now possible to exploit large datasets and to refine analytical procedures in support of editorial decisions that a traditional edition could not accommodate. In the context of early music – from the study of the Ars antiqua through the XVIe century – numerous issues inherent to musical notation arise, such as the faithful transcription of ligatures in notation software, which often struggles to capture the precise subtleties of handwritten calligraphy. The availability of a digitized source therefore helps to overcome these difficulties by combining graphical fidelity with the interoperability of encoding. One relevant approach could involve designing synoptic editions, creating direct visual links between manuscripts and recordings, thereby facilitating the interpretation of tripla and mensural proportions. It is essential to address the question of access to sources for users of digital editions to enable a better understanding of historical practices, particularly regarding prosody and musical arrangements. Indeed, while some variants may be considered negligible, others are crucial for interpretation. Access to sources 1 X. Fresquet (December 4, 2022). IIIF and the Connectivity of Medieval Mediterranean Music Collections. MnemoMed. Retrieved February 6, 2025 from https:// doi.org/10.58079/vwx2 is generally possible through two main avenues : a link to a resource hosted by an institution that has previously worked on or directly digitized the source, or through an IIIF viewer integrated directly into the digital edition. It is nevertheless important to consider the legal implications of such decisions, taking into account the location and institutional ownership of the sources, as well as, where applicable, pre-existing digitizations, which must be properly referenced. The Polish Music Heritage project offers an ideal model : connections between digital scores and digitized sources are explored using IIIF. It is thus possible to zoom in on the manuscript maintaining excellent image quality, and easily access all the information related to the document’s dimensions and format, accompanied by a color scale to verify hues (fig. 22). The interface is user friendly and compatible with diverse formats (PDF, MEI, Humdrum, etc.). 34 3. design fig. 22: sza na Cztery głosy by Wincenty Studziński and Piotr — digitization of the score1 and visualization of the transcription in MEI.2 3.4. Data Lifecycle, Management, Accessibility, and Reuse The management of the data lifecycle is a central component of institutional data management policies. It aims to oversee all stages – from creation to archiving or destruction – while ensuring data quality, security, and accessibility. From the moment of their creation, data must be 1 This source can be consulted via this link : https://polish.musicsources.pl/en/lokalizacje/galeria/rekopisy/6083-msza/3 2 The encoding can be consulted via this link : https://polish.musicsources.pl/en/lokalizacje/11-archives-and-library-of-cracow-cathedral-chapter 3 Our translation of https://data.unige.ch/cycle-de-vie/etapes-du-cycle-de-vie documented and stored according to institutional and legal standards to guarantee their integrity and potential for reuse. During their use, governance mechanisms should enable control over their usage and sharing in compliance with regulatory and legal frameworks, particularly concerning the protection of sensitive data. [...]3 35 3. design 3.4.1. Digital Edition or Database Given the power and comprehensiveness of the tools now available, the boundary between a digital edition equipped with a critical apparatus and a database must be examined. Indeed, the introduction of an excessive number of metadata fields and crossreferences can sometimes shift the musical paradigm toward a knowledge organization system with a historiographical focus, thereby diluting the primary information. It is therefore important to ensure that a project’s feasibility is not compromised by overly ambitious goals. From the earliest stages of an editorial project, it is necessary to question which fields and crossreferences are absolutely necessary to initiate the creation of an Open Access project. Should priority be given to the strictly musical and analytical study of a composer’s works, or rather to their historiography, the locations of their performances, and their reception history ? While the former clearly falls within the scope of a critical music edition, the latter would benefit far more from a database architecture, whether relational or semantic. Dezède stands out among databases that offer considerable richness, advanced functionality, and a the opportunity for collaboration : this opensource web application is specifically designed to archive performances and put them in chronological order. Maintained by an active community, it “fosters the development of research on the circulation of works and performers, by providing researchers and the public with a wealth of historical data.”1 1 https://dezede.org/presentation 2 The graph is interactive and allows users to visualize which composers were most frequently performed together : https://dezede.org/dossiers/concerts-abbaye-de-royaumont/stats fig. 22: Chordal graph representation of the most performed composers at Royaumont Abbey between 1936 and 19772 (shown alongside). 36 3. design These issues particularly affect digital editions in MEI format. Given the format’s powerful interoperability, it can be tempting to overload files with metadata and references that exceed the project’s original objectives and complicate the editorial process. It is therefore important to remain attentive to these issues so that the edition remains, above all, an effective tool for its users and can be easily updated, a task that is made easier when the project’s technical architecture is planned from the outset. 3.4.2. Data Sustainability and Visibility Although many secure local alternatives exist, data storage and hosting on servers remains costly. Virtual machines1 can be provided by universities or large research infrastructures (notably the IR* HumaNum for the Humanities and Social Sciences). Storing data on national servers offers numerous advantages, particularly for publicly funded and Open Access projects, but also presents disadvantages in more complex geopolitical situations. The French government has recently proposed a list of secure repositories for the humanities and social sciences,2 including Nakala, which is used for projects associated with IR* HumaNum. It is a “data repository that aims to preserve and disseminate data produced by French research projects in the Humanities and Social Sciences, in compliance with the FAIR principles” and “is primarily intended for research projects carried out by institutions affiliated with the French 1 A virtual machine is a digital version of a physical computer, stored on and running from a physical machine. It can host an operating system and, consequently, databases, as well as perform computations and other tasks. 2 https://recherche.data.gouv.fr/fr/entrepots 3 https://www.nakala.fr/about 4 https://neuma.huma-num.fr/ Ministry of Higher Education and Research.”3 In a similar spirit – also hosted by HumaNum but focused specifically on musical content – NEUMA can be cited as “a digital library that gives on-line access to rare and often hardly accessible corpora. […]”4 The final stage of an edition thus involves making a well-considered and appropriate choice for stable and highquality hosting of data, combined with a userfriendly consultation interface. 37 4. conclusion 4.1. Checklist for Digital Editions; What are the Essential Steps ? Having outlined the main issues associated with editing music using digital tools, readers of this guide should now be prepared to begin a first edition of a short work. The checklist below provides a foundation for beginning the editorial process, ensuring that all the main steps are observed : $ Is my corpus clearly defined ? $ Do I need to reconstruct the work, or is it complete ? $ Have I decided to produce a diplomatic edition or a critical edition ? $ If I have chosen to prepare a critical edition, do I have access to all existing editions ? $ If I digitized the sources myself, did I respect IIIF protocol ? $ Have I clearly defined the intended audience for my edition? Students, the musical community, and/or musicologists ? $ Does my project adhere to the budget and schedule allocated to it ? $ Have I considered the expected dissemination model– Open Access or subscription-based format ? $ Have I identified the encoding software1 I will use as well as the formats Iwish to employ for export2 ? Once my file has been exported, I must ensure that metadata are specified precisely: $ Have I included all essential metadata, ensured interoperability, and linked the work to its RISM ID ? $ Have I cited my sources ? 1 In early 2025, in light of indications that the development of Sibelius may no longer be maintained, we recommend using MuseScore, which provides numerous advantages through its Open-Source design and its easy compatibility with the MEI format. 2 Given its interoperability, flexibility, and easy export from MuseScore, we recommend using MEI format, whose visualization is fully supported by Verovio. $ Have I acknowledged all contributors to the edition ? $ Assigned a license (Creative Commons, etc.) to my work ? Once the file metadata are complete, the next step is visualization: $ Am I satisfied with the quality of the rendering in a viewer ? $ Have I confirmed that the file has been validated in MEIfriend ? $ Am I satisfied with the quality of my encoding for the purpose of analysis? $ Have I defined a clear strategy for annotating variants? $ Is the user interface sufficiently clear ? $ Have I identified a trusted repository for depositing my data ? $ Have I ensured that my edition is correctly indexed and referenced ? $ $