Computer Assisted Music Theory / Analysis ?

[DrFaustus666]

Does anyone know which universities (if any) are top-ranked in automated or computer-assisted analysis of music?

I am pipe-dreaming of writing computer programs to perform analysis, but I am wondering if

i) it's already old hat; or

ii) it's so far-fetched that I'd be laughed at should I put this in a statement of purpose; or

iii) some other situation exists that I can't imagine.

Thanks for any and all comments.

John

[jose]

The only one I know well is Columbia university - they're big on spectral analysis of sound.

I think Stanford is also really good, and maybe Princeton (at least, princeton used to be... not sure what the deal is nowadays).

Edit: I think if you're talking about tonal analysis of music (ie: what humans can do) then it is a pretty outdated thing, but in terms of analyzing sound and acoustics, there's a lot going on at Columbia and it's still quite cutting edge, imho.

Edited February 18, 2010 by jose

[DrFaustus666]

The only one I know well is Columbia university - they're big on spectral analysis of sound.

I think Stanford is also really good, and maybe Princeton (at least, princeton used to be... not sure what the deal is nowadays).

Edit: I think if you're talking about tonal analysis of music (ie: what humans can do) then it is a pretty outdated thing, but in terms of analyzing sound and acoustics, there's a lot going on at Columbia and it's still quite cutting edge, imho.

Hi Jose,

Thank you! Yes, indeed I was talking about tonal analysis ... and I imagined it's probably either outdated, or technology is nowhere close yet. Too bad. Hmmmm.... sound and acoustics, eh? That's worth looking into.

Again, thanks for your response.

John

[jose]

I don't know too much about tonal analysis software - I remember hearing about programs that analyze large amounts of certain highly stylized kinds of music (Bach Chorales, for example), I believe they then extrapolate algorithms from these analyses and use them as a way to automatically "compose" in the style. I've been trying to remember some of the people/programs but can't really recall.

This is a well-known sound analysis program that a Columbia grad student developed some time ago... that's more the kind of thing they're doing over there and in Stanford, I think.

[DrFaustus666]

I don't know too much about tonal analysis software - I remember hearing about programs that analyze large amounts of certain highly stylized kinds of music (Bach Chorales, for example), I believe they then extrapolate algorithms from these analyses and use them as a way to automatically "compose" in the style. I've been trying to remember some of the people/programs but can't really recall.

This is a well-known sound analysis program that a Columbia grad student developed some time ago... that's more the kind of thing they're doing over there and in Stanford, I think.

Thank you again Jose,

I checked out the URL and will look at it in depth in the next week or two. I'm not a composer, or rather, all my attempts at composition have been dismally derivative. (Although, as an ex-trombonist and once-decent orchestrator, I do still have in the back of my head the dream of writing a really good trombone concerto, that would be first-rate music, but still technically within the reach of the average college orchestra and student soloist.)

The kind of thing I originally had in mind was the Bach Chorale analysis you referred to. Though I had more ambitious plans.

I'd like to write a program that would analyze a wide variety of music, looking for patterns, melodic, harmonic, structural, etc: the idea being to find heretofore undiscovered relationships between . . . God knows what? "The influence of Ravel on The Beatles" or "Tropes of Gregorian chants in the music of Elliot Carter".

What I'd really like to know is "Has that ground already been covered?" If so, then I'd like to head into less well charted seas (yes I know I mixed metaphors ).

Cheers,

John

[jose]

That would be quite a feat if you could write something like that!

I think the issue with that kind of deep analysis is that it's highly subjective, you get theorists arguing for and against the same thing, a lot of times based on intuition or how they personally hear/perceive X harmony or note in a context. The same note or harmony can often be correctly analyzed in many different ways. The difficulty would be setting up the program so it doesn't simply do a surface analysis, in which case totally random superficial similarities between disparate works could be misinterpreted as some kind of fundamental similarity between them.

Schenkerian analysis is the traditional "deep structure" analysis that tonal (and I guess even some post-tonal) analysts use to determine very broad underlying structure of works... that could be something you include in the program to analyze tonal content and direction of a work's form.

[DrFaustus666]

That would be quite a feat if you could write something like that!

I think the issue with that kind of deep analysis is that it's highly subjective, you get theorists arguing for and against the same thing, a lot of times based on intuition or how they personally hear/perceive X harmony or note in a context. The same note or harmony can often be correctly analyzed in many different ways. The difficulty would be setting up the program so it doesn't simply do a surface analysis, in which case totally random superficial similarities between disparate works could be misinterpreted as some kind of fundamental similarity between them.

Schenkerian analysis is the traditional "deep structure" analysis that tonal (and I guess even some post-tonal) analysts use to determine very broad underlying structure of works... that could be something you include in the program to analyze tonal content and direction of a work's form.

Yes, Schenkerian analysis is exactly what I had in mind. And yes, I know it's probably a 10-year project to automate --- but I'm a working programmer with many years under my belt, and will soon be of retirement age, and I want to do something really noteworthy before I kick off this mortal coil.

John

[jose]

Best of luck to you John!

[DrFaustus666]

Best of luck to you John!

Thanks, and to you too!

[liszt85]

I don't know too much about tonal analysis software - I remember hearing about programs that analyze large amounts of certain highly stylized kinds of music (Bach Chorales, for example), I believe they then extrapolate algorithms from these analyses and use them as a way to automatically "compose" in the style. I've been trying to remember some of the people/programs but can't really recall.

you mean this?

Also read: Larson, Steve (2001).Dear Emmy: A Counterpoint Teacher's Thoughts on EMI's Two-Part Inventions. In David Cope, Virtual Music: Computer Synthesis of Musical Style. MIT Press, 237-262.