May 10 CIRMMT Workshop
Posted by Catherine Motuz on July 07, 2013
On Friday, May 10th and Saturday, May 11th, the ELVIS team met here at McGill to exchange ideas and to work together on pushing ELVIS forward.
The session began on Friday with a series of talks by each of the teams to a crowded seminar room, showing what directions they have been following in their research. Julie Cumming, Catherine Motuz and Christopher Antila spoke on behalf of McGill, showing the ELVIS database and especially the VIS software that we have been using to analyze vertical interval successions. The team from Yale, including Chris Whyte (who, replacing his advisor who couldn’t make it for the conference, dubbed himself “the poor man’s Ian Quinn”) and Kirill Zikanov, explained the projects that they are working on. These involved the classical-archives.com database of over four thousand pieces in MIDI format, including Byrd, Vivaldi, and major composers from Bach to Wagner. The main metadata (composer, date etc.) has all been added by hand, and Chris and Kirill are working on adding chord data and on key-finding algorithms. The chord data can be used to plot style similarity from one composer to the next, for instance by analyzing all of the harmonic progressions of one composer’s corpus and seeing how well it describes the corpus of other composers (eg. how well does Handel’s model describe Bach? Schumann? Debussy?).
Relying more heavily on that of the Josquin Research Project, they have also developed a system for finding the metre of pieces, both by detecting which note value carries the “beat” and by calculating which beat corresponds to an upbeat or downbeat. On certain repertoires, their algorithms work astoundingly well, and they are working to expand the success of their method to early music by implementing a suspension-aware beat-strength algorithm to work with the others.
Frauke Jürgensen and Ian Knopke then presented on behalf of the team from Aberdeen. Frauke addressed the issue of in what temperament one should tune an organ to play from the Buxheim organ book. She tested with computers Mark Lindley’s hypothesis that the pitches in the manuscript suggest meantone rather than pythagorean tuning, and her statistical analysis confirmed his findings even when elaborated to include the metrical position of triads and relative duration triads to one another.
Ian then presented his work on finding contrapuntal modules that recur in multiple pieces within a single composer’s corpus, in this case Palestrina’s. Using a single letter to represent every pitch and interval, he then put the letters into a suffix tree. Then, employing searches developed in the field of bio-informatics (gene sequencing in particular), he can study the entropy of particular sets.
Frauke then presented a study on “re-booting” the ears of modern listeners. Listeners of the late sixteenth century found the music of Monteverdi to be shocking, but to modern ears it is quite pleasing in comparison to some later musics. Frauke exposed groups of students to various repertoires before having them listen to some of Monteverdi’s most controversial (for their time) compositions, allowing them with some success to experience the composer’s works with sixteenth-century ears. The main focus of her talk, however, was in addressing why Monteverdi’s music was considered shocking. By comparing the music of Monteverdi to that of Palestrina by counting the number of dissonances in each composer’s works, she found that Palestrina was found to be in fact more dissonant, so the controversy of Monteverdi’s works must lie in a different musical parameter. The Aberdeen contingent then concluded their presentation by asserting the need for more ground truth to confirm or refute results produced by computer analyses.
Myke Cuthbert finished the series of talks with an overview of four projects he is currently working on. The first involved a presentation of Music21, emphasising the advantages of a built-in corpus and of its modular expansion capability. He then outlined the EMMSAP: Electronic Medieval Music Score Archive Project (1300-1420), through which 300 pages of polyphonic music of the fourteenth century are being digitally encoded, partially by expert encoders and partially using optical music recognition. He then moved on to his work directly with ELVIS, using MAPReduce to index tasks so that multiple machines can work together to solve corpus queries in one hundredth of the time that it would normally take. Finally, he outlined his goals for the future, which include making harmonic reductions of 14th-century pieces, weighing sonorities by their position in a measure, and reducing arpeggiated chords to vertical ones.
After the talks, the student contingent made their way to Benelux to continue chatting on an informal level, and in the evening, we all convened for excellent Indian food at nearby Asha.