Wednesday 3 November 2021, 4.00PM to 5:30pm
Despite the advances in technology which have enabled the increasing acceptance of the Digital Humanities in recent years, musicology has been hampered by the relative paucity of useable data - a condition cited as a significant problem by Nicholas Cook over 15 years ago. DH can, of course, make immediate use of textual and numerical data, which exists in vast quantities in publicly accessible digital repositories. Music, on the other hand, is manifest on the web mostly in the form of audio recordings which always need sophisticated digital analysis before they can be treated as sources of data for analysis. For the study of musical performance, recordings are of course essential, and it is true that they can be accurately aligned with a suitably encoded score, if one exists. But there still exists an inevitable gap between a musical score and an audio recording made from it which limits the extent to which what I here call ‘real musicology’ can be done this way.
To narrow this gap, another approach, perhaps somewhat closer to traditional (non-digital) musicology, is to start from digital images of the scores and process them with optical music recognition (OMR) software; in principle, this automatically generates digital encodings for further processing and analysis. However, as everyone who has used OMR knows, even the best OMR programs generate errors unless the score in question is expertly printed and, furthermore, is trivially simple, in perfect physical condition and photographed perfectly. What is more, for historical scores (which may not follow the conventions of modern engraving) specialist software is needed to deal with unfamiliar musical symbols and layout, not to mention other notation types such as tablatures. But given such software, large quantities of data (albeit error-prone) can be produced from digital images such as the PDFs of historical printed music available on IMSLP.
My talk will focus on my own work in the field of early printed music aiming to produce useable tools for information retrieval while sidestepping the problems caused by OMR error. By ‘useable’, I mean fast enough to allow a user to explore a collection without waiting more than a second or so - much as we search the web with Google. I’ll be talking about my pilot work on the Early Music Online collection, and its more recent enhanced version, F-TEMPO (Full-Text search of Early Music Online), which currently has about half-a-million pages in its index, including all early typeset printed music (16-17th centuries) from the Bavarian State Library in Munich. Although just a few startling discoveries were made during development, these give the hope that many more could emerge from systematic investigations of the collections we have been able to index.
Attend the seminar in person, or online via zoom (Meeting ID: 998 9220 5308; Passcode: 390939).
Postponed from October.
Location: In-person at Sally Baldwin Buildings D Block (Dept of Music), Room 003, and online via Zoom