[Smt-talk] RN analyzers

Phillip Kirlin pkirlin at cs.umass.edu
Wed Feb 22 11:09:15 PST 2012


I do not know any complete software packages that will do this, but I 
can give you some pointers to where there might be one lurking.

I have read many papers about this problem and lots of approaches break 
this down into two phases: first analyze the music to find chord labels 
(in the sense of G major triad, D dominant seventh chord, etc) and where 
they change, and then turn the labels into Roman numerals.  I believe 
you will find much more material on the first phase than the second 
phase, since I think the second phase turns out to be harder.

Here are some resources:

David Temperley. 2009. "A Unified Probabilistic Model of Polyphonic 
Music Analysis." Journal of New Music Research 38, 3-18.  (This has 
downloadable software, but it will only give you chord labels, no Roman 

C. Raphael and J. Stoddard Harmonic Analysis with Probabilistic 
Graphical Models (ISMIR 2003) Computer Music Journal vol 28:3, 45--52, 
2004.  (again, only score to chord labels, no Roman numerals)

B. Pardo and W. Birmingham "Algorithms for Chordal Analysis" 	Computer 
Music Journal 	vol. 26 (2), pp. 27-49, 2002.  (no Roman numerals)

W. Bas de Haas, HarmTrace: Automatic functional harmonic analysis. 
Technical Report UU-CS-2011-023 July 2011
Department of Information and Computing Sciences
Utrecht University, Utrecht, The Netherlands (this goes from chord 
labels to Roman numerals)

All of these are available online.

I can probably come up with some more links, but the short answer is 
that I don't know of a system that goes straight from score to Roman 
numerals.  I know many people have evaluated systems using the 
Kostka-Payne corpus [2] which does have Roman numerals.  I think the 
KernScores or MuseData websites may have some data that's annotated with 
Roman numerals, but I can't seem to find it at the moment.

Incidentally, I built one myself a few years back (I think it went from 
musicxml to Roman numerals) but it wasn't especially intelligent (for 
example, I don't think it distinguished between cadential six-fours and 
"true" I(6/4) chords).

I'd be happy to talk more about this with you or your student, since my 
area of research is computational music analysis, and I'm always looking 
for more people who are into this stuff.

[1] http://www.ismir.net
[2] http://theory.esm.rochester.edu/temperley/kp-stats/index.html


Phillip Kirlin
Department of Computer Science
University of Massachusetts

On 2/22/12 9:19 AM, Dmitri Tymoczko wrote:
> Are there any computer programs out there capable of inputting a generic XML score of a classical-style keyboard piece (or string quartet or what have you) and outputting a Roman numeral analysis?
> I ask because I have a student who is built one, and we're looking to test its accuracy against the best things out there.
> Thanks,
> DT
> Dmitri Tymoczko
> Associate Professor of Music
> 310 Woolworth Center
> Princeton, NJ 08544-1007
> (609) 258-4255 (ph), (609) 258-6793 (fax)
> http://dmitri.tymoczko.com
> _______________________________________________
> Smt-talk mailing list
> Smt-talk at lists.societymusictheory.org
> http://lists.societymusictheory.org/listinfo.cgi/smt-talk-societymusictheory.org

More information about the Smt-talk mailing list