The Music Encoding Initative (MEI), started as a ‘one man show’ at the University of Virginia. Perry Roland first designed an XML based encoding scheme for representing music notation according to principles and concepts that the Text Encoding Initiative was putting into practice for scholarly text encoding (Kepper, 2010). This effort addressed academic encoding approaches since the beginning, thus attracting a group of scholars interested in building a community around the MEI specification. In 2007, the format underwent its first cooperative improvements that added support for medieval notation, editorial interventions, and alignment with facsimile.
These personal initiatives prompted further involvement and soon could be channeled into the formation of a MEI study group in 2009. Since then, institutional support and public funding helped improving MEI. A project funded by the NEH and DFG led to the first community effort by making the first non-beta release of MEI publicly available in May 2010 (Röwenstrunk 2010). To date, the MEI has developed into an open and community-driven effort involving a council, a technical team and a growing user group.
This emerging community intends to improve the involvement of musicology within the Digital Humanities field. With this aim, some members have previously introduced examples of MEI usage to the DH community (see Kepper 2010; Viglianti 2010). Today, MEI’s users are mostly involved in projects that employ the format and/or develop software to support MEI use. This poster intends to demonstrate some of these efforts, particularly concerning the tool support for day-to-day work of the scholar for entering music notation and metadata, creating facsimile alignments and for the production of digital editions.
Transcribing or entering music notation with a computer is typically a more laborious task than transcribing text; this is because, unlike letters, music symbols themselves require complex codes to be represented. Many machine-readable formats, such as Humdrum or Lilypond, rely on ASCII-based structures which can be more or less intuitive to type. However, the most common note entry is done with WYSIWYG score editors, which allow the transcriber to enter the notation directly on a virtual score.
Note entry also lies at the heart of encoding music documents with MEI; however, encoding music notation in XML immediately poses a problem of overlapping hierarchies, because the score is organized as a grid that represents temporal sequence horizontally (one event comes after the other, such as a sequence of notes) and temporal co-occurrence vertically (events occur at the same time, such as a chord on the piano). The general complexity of music codes and the unavoidable workarounds for dealing with this condition makes hand-encoding a strenuous task.
Given these premises, it is evident that tools are necessary to enter music notation into the MEI format in a more intuitive manner. At this stage of MEI development as a community, there are some options available for simplifying note entry.
1. WYSIWYG score editors. In the past ten years, MusicXML has had outstanding success as an interchange format between different score editors. As a consequence, most specialized software now provide an export to MusicXML. The MEI provides XSLT stylesheets to convert from MusicXML 1.0 and 2.0 to MEI, which makes it possible to use score editors’ exports to be converted into MEI. This, nonetheless, is not necessarily a straight-forward process. Given the graphical, non-linear organization of notation on the page, each score editor can be more or less precise about the graphical positioning of symbols on the page. Symbols such as directions, phrase marks, and dynamics are particularly affected as they can be attached either to a staff, a measure, or a specific event on the staff. These differences are seemingly innocuous to the score editor’s users because their rendition on the page looks the same. However, when moving into a semantic representation system such as MEI, the associations of different symbols and their start/end points matter. Notwithstanding these difficulties, using score editors and MusicXML still simplifies note entry in MEI. Some post-processing, however, is still necessary.
2. MEI-specific score editors. As described above, MEI has existed for more than ten years, but has experienced a considerable community effort only in more recent times. Amongst several uplifting activities, a few focused on the creation of WYSIWYG editors able to export directly into MEI. The TextGrid component MEI Score Editor (MEISE) is the most complete to date.1 Besides exporting notation directly to MEI, it is designed to be able to represent editorial aspects supported by the format, such as variants and alternative readings.
3. XML editors. The first two solutions may be sufficient for most encoding scenarios, especially if MEISE grows to be more comprehensive and/or is followed by other open source efforts for creating MEI exports for more common score editors. It is possible to imagine, however, that MEI will be used for deep-encoding projects, such as critical editions, diplomatic or genetic transcriptions, analyses, etc. For such scenarios, it is debatable that a generic tool would be able to cover all uses. Obtained as an heritage from Text Encoding Initiative, MEI’s flexibility is both a blessing and a curse. It allows one to use the format as a framework within which it is possible to build specific encoding models through customizations; at the same time, though, it makes it impossible to build tools that can cater for all possible declinations of the format.2 Nonetheless, tool development remains an essential component of the MEI’s community efforts; particularly towards building tools tailored to specific tasks or in the form of libraries for manipulating MEI documents.3 Those project that intend to use MEI at its full power, however, will eventually need to resort to the use of XML editors. Editing the XML manually is doable with a little training, but requires some imagination to see the score through the code. Alternatively, one could exploit many modern XML editors’ visual support for text encoding. This is typically done through CSS-based previews that can be edited in real-time. A similar scenario is considerably harder to obtain with music notation; however experimentation is underway.
MEI as a format was – amongst other goals – initially intended to ‘represent the common expressive features of traditional facsimile, critical, and performance editions’,4 which opens the way for more and more critical edition projects to describe, preserve and publish their work with MEI.5 Because of the very small or even non existing support of score editors or renderers in the last years, most endeavors to create and publish digital music editions have been based on facsimiles and texts. Other formats could not bridge the gap, because of their limited representation of textual phenomena especially in the fields of editing and transcription.
With the latest possibilities for note entry and score rendering, digital editions are going to reach a higher level of quality. For example, searching and analysing music editions would become possible, similarly to what already happens in digital editions of literature. Moreover, readability and usability would be improved by selectively rendering scores in modern or old notation or in different keys.
The Edirom project started in 2006 with the aim to provide tools for creating and presenting historico-critical music editions. The project was one of the main contributors in moving MEI to a community-driven standard.6 Given the lack of score writers for MEI, the Edirom tools kept using image based presentation techniques. But with emerging tools for MEI, more and more encoded music is being included for different purposes. Recent efforts include a score renderer by Thomas Weber which uses the web standard SVG to enable digital presentations with Edirom tools to show ad hoc generated notation and allow interaction with musical snippets like switching versions or highlighting specific symbols. Eventually, editions will no longer have to be constrained to providing one single edited text for scholars;7 not every uncertainty will have to be resolved; and equitable versions could be presented simultaneously. Hence, the user might have a closer look at the underlying editorial work (see Bohl, Kepper & Röwenstrunk 2011).
In order to provide these functionality and level of knowledge, one needs a detailed encoding of music documents as described above, and a considerable amount of interlinking. Such data may then be used in the Edirom Editor for creating digital music editions. The Editor, for example, provides: a) mechanisms to link structural, musical information (like a movement or measure) to regions on facsimiles; b) functions for formally describing relationships or differences between these music documents. Some of the specific tasks in the workflow of creating digital editions are handled by specific tools outside or included within the Edirom Editor; such as semi-automated recognition of measure positions with Optical Music Recognition techniques or describing the metadata of a document, which is done with a included version of the Metadata Editor and Repository for MEI Data (MerMEId) from the Danish Centre for Music Publication.8
Bohl, B. W., J. Kepper, and D. Röwenstrunk (2011). Perspektiven digitaler Editionen aus Sicht des Edirom-Projekts. Die Tonkust 5.
Dabbert, J., and J. Veit (2010). MEISE: an editor for encoded/encoding music. TextGrid Advisory Board Meeting, Detmold. Available at https://www.textgrid.de/fileadmin/TextGrid/konferenzen_vortraege/fachbeirat_0710/meise.pdf (accessed November 2011).
Kepper, J. (2010). A Data Model for Digital Musicology and its Current State – The Music Encoding Initiative. Digital Humanities 2010, London. Available at http://dh2010.cch.kcl.ac.uk/academic-programme/abstracts/papers/html/ab-817.html (accessed November 2011).
Pierazzo, E. (2010). Editorial teamwork in a digital environment: the edition of the correspondence of Giacomo Puccini. Jahrbuch für Computerphilologie 10.
Röwenstrunk, D. (2010). Digital Music Notation Data Model and Prototype Delivery System. Forum Musikbibliothek 31.
Viglianti, R. 2010. Critical Editing of Music in the Digital Medium: an Experiment in MEI. Digital Humanities 2010, London. Available at http://dh2010.cch.kcl.ac.uk/academic-programme/abstracts/papers/html/ab-819.html (accessed November 2011).
1.A first beta release of MEISE is to be made available by the end of 2011.
2.‘[H]umanities data have a very high rate of variation [...], and even when similarities do exist, the differences are important enough to imply that designing tools able to cope with such variety would be a very demanding task’ (Pierazzo 2009).
3.See for example libmei http://ddmal.music.mcgill.ca/libmei
5.The german projects OPERA (http://www.opera.adwmainz.de/), Reger-Werkausgabe (RWA, http://www1.karlsruhe.de/Kultur/Max-Reger-Institut/de/rwa.php) and Carl-Maria-von-Weber-Gesamtausgabe (http://www.weber-gesamtausgabe.de) are using the tools provided by the Edirom project.
6.The project organized a conference Digitale Edition zwischen Experiment und Standardisierung, 2007 in Paderborn (Germany), where Perry Roland was invited to present MEI. Later, members of the project co-applied for funding for organizing two more conferences where the MEI council was constituted (see Röwenstrunk 2010).
7.There will probably always be a normalized version of the edition for practical use.