sandell@ferret.ils.nwu.edu (Greg Sandell) (10/18/89)
In article <5013@orca.WV.TEK.COM> steveb@eve.WV.TEK.COM () writes: >I have long been contemplating writing software that can analyze tonal music >via Schenker techniques. Every time I attempt to scope out the problem I seem >to increase in complexity by an order of magnitude. I am looking for folks who >have attempted to tackle this problem. Specifically I am looking for >collaborators but I would be happy to start with references in the literature. >I am also looking for information on potential file formats and encoding >schemes for the input side. Any information would be extremely helpful. > >Steve >------------------------------------------------------------------------------- >FROM: STEVEN C. BILOW -- Software Engineer, Tektronix >EMAIL: steveb@orca.WV.TEK.COM PHONE: (503) 685-2463 >USMAIL: P.O. Box 1000 61-028, Wilsonville, OR 97070-1000 Stephen Smoliar, no stranger to the net (esp. rec.music.classical) has an article titled "A computer aid for Schenkerian analysis," COMPUTER MUSIC JOURNAL 4/2, 41-59. Another item is by James Snell (1979), "Design for a formal system for deriving tonal music," but you will find it practically impossible to find since it is a Master's thesis (State University of New York at Binghamton). Lerdahl & Jackendoff's A GENERATIVE THEORY OF TONAL MUSIC (MIT Press, 1983) contains a hierarchical theory of music which is Schenker-like in flavor. One particularly nice thing about the approach is that the rules are fairly explicit, which lends itself more to a computational treatment than raw Schenker theory. I think the problem is well worth working on, although as you say, it will be very complex. The fact that music theorists frequently disagree as to what consititutes a `correct' Schenkerian analysis shows that much of the theory is arcane and implicit. (By the way, you can always tell if someone is a truely devoted follower of Schenker if they use the word `correct' alot in their writings.) My advice, for what it's worth, is to pick a very simple musical domain (e.g. early 18th century minuets) and try to acheive success there first. Good luck, Greg Sandell *************************************************************** * Greg Sandell, Institute for Learning Sciences, Evanston, IL * * sandell@ferret.ils.nwu.edu * ***************************************************************
Q2816@pucc.Princeton.EDU (Creative Business Decisions) (10/19/89)
In article <1325@accuvax.nwu.edu>, sandell@ferret.ils.nwu.edu (Greg Sandell) writes: >In article <5013@orca.WV.TEK.COM> steveb@eve.WV.TEK.COM () writes: >>I have long been contemplating writing software that can analyze tonal music >>via Schenker techniques. Every time I attempt to scope out the problem I seem >>to increase in complexity by an order of magnitude. I am looking for folks who >>have attempted to tackle this problem. Specifically I am looking for >>collaborators but I would be happy to start with references in the literature. >>I am also looking for information on potential file formats and encoding >>schemes for the input side. Any information would be extremely helpful. >Stephen Smoliar, no stranger to the net (esp. rec.music.classical) has an >article titled "A computer aid for Schenkerian analysis," COMPUTER MUSIC >JOURNAL 4/2, 41-59. Another item is by James >Snell (1979), "Design for a formal system for deriving tonal music," >but you will find it practically impossible to find since it is a >Master's thesis (State University of New York at Binghamton). >Lerdahl & Jackendoff's A GENERATIVE THEORY OF TONAL MUSIC (MIT Press, [etc.] Also Michael Kassler's Princeton Ph.D. thesis. Complete reference on request, or just call UMI's 800 number. >I think the problem is well worth working on, although as you say, it >will be very complex. The fact that music theorists frequently >disagree as to what consititutes a `correct' Schenkerian analysis >shows that much of the theory is arcane and implicit. (By the way, >you can always tell if someone is a truely devoted follower of >Schenker if they use the word `correct' alot in their writings.) >My advice, for what it's worth, is to pick a very simple musical >domain (e.g. early 18th century minuets) and try to acheive >success there first. I don't believe there IS such a thing as a "correct" Schenkerian analysis, and I consider myself something of a Schenkerian. You see, most Schenkerian analyses are done in order to show something SPECIFIC about the INDIVIDUAL piece. The "point" of the analysis determines the way in which it is done. Schenkerian analysis is NOT a game of "Find the Ursatz." Often, the purpose is to demonstrate motivic relationships between structural levels, to demonstrate the relationships between sections of a larger piece, or for another reason. Just "analyzing" in a rote fashion tends to get you nowhere, as Karl-Otto Pluem's work shows. (He did a dissertation by doing analyses of all the Bach 'cello suites, I think.) In general, the best analyses are based on an intuitive "hook" of some sort. Some of the best motivic-Schenkerian analyses were done by Ernst Oster. Two gems can be found in reprint at the end of _Aspects of Schenkerian Theory_, ed. David Beach. New Haven, Yale UP, ca. 1983. One of them, of the Chopin Fantasy-Impromptu, comes up with results no computer is capable of. (The word "intertextuality" is appropriate...) The other one is a bit more conventional, but comes up with motivic relationships of a very subtle nature. (Egmont Overture.) These analyses TELL you something about the piece in its uniqueness. I don't think reduction or generation rules alone will get you those results. Another thing to consider is the historical aspect of music. What works do your rule work for? What does variance from the rules say about a given work? Also, how will you integrate the text of a vocal work into your analysis? Its (often ambiguous) structure and meaning? Finally, though I like Schenker's method and many of the analyses it has produced, I do not necessarily accept reducability as a condition of well-formedness; at least, I believe that there is always some ambiguity in the way great works are heard, even when one is hearing them for the nth time. A good analysis can point out these ambiguities; how hard will it be to get your program to do so? In short, ask yourself the following questions: What is Schenkerian analysis? What is it used for? What does one learn from an analysis? Why do *I* want to do it? And what will the computer do for me? Roger Lustig (Q2816@PUCC.BITNET Q2816@pucc.princeton.edu) Disclaimer: "That is not my dog."
smoliar@vaxa.isi.edu (Stephen Smoliar) (10/19/89)
In article <1325@accuvax.nwu.edu> sandell@ferret (Greg Sandell) writes: >In article <5013@orca.WV.TEK.COM> steveb@eve.WV.TEK.COM () writes: >>I have long been contemplating writing software that can analyze tonal music >>via Schenker techniques. Every time I attempt to scope out the problem I >>seem >>to increase in complexity by an order of magnitude. I am looking for folks >>who >>have attempted to tackle this problem. > >Stephen Smoliar, no stranger to the net (esp. rec.music.classical) has an >article titled "A computer aid for Schenkerian analysis," COMPUTER MUSIC >JOURNAL 4/2, 41-59. Another item is by James >Snell (1979), "Design for a formal system for deriving tonal music," >but you will find it practically impossible to find since it is a >Master's thesis (State University of New York at Binghamton). > First of all, I would like to thank Greg for cross-posting this to rec.music.classical, thus allowing me to get in on this discussion. Secondly, I should point out that the "computer aid" discussed in the aforementioned article is basically a structure editor (as was observed by Jim Meehan in HIS article in that same issue of COMPUTER MUSIC JOURNAL). Thus, it does not offer very much as far as automating the process of analysis. Rather, it provides a consistent structural foundation for what those graphs are REALLY trying to say (i.e. a hierarchy of elaborations). Thus, it is an aid for someone who has already conceived of the analysis and wishes to make sure it is expressed consistently. My first step away from this direction is recorded in the Report of the Twelfth Congress of the International Musicological Society, in which I had to respond to a paper by Michael Kassler. At that point I was beginning to realize that there was more to music than any sort of "parse tree" representation such as could be offered by Schenker graphs. I felt that music analysis had to follow the lead of natural language processing and start looking for ways to represent semantics (whatever that might mean) as well as syntax (i.e. well-formedness). This was the beginning of my migration towards issues of memory, such as those I am now trying to develop. It should be relatively apparent that I am not really a member of the Schenker camp. Thus, I reject this whole idea of "correctness" which Greg cited. I believe that the Schenker notation is a good step towards documenting a particular performance of a composition (to the extent that it distinguishes structural from auxiliary notes). However, I feel strongly that there may be two quite different interpretations which are equally "correct." Thus, if you want a computer to do its own analysis, it is worth asking why you wish to undertake the task. One reason might be for the machine to advice you on how to play it. However, it is unlikely that a machine will ever give you a "definitive" analysis of a composition, since it is unclear that such an analysis exists. ========================================================================= USPS: Stephen Smoliar USC Information Sciences Institute 4676 Admiralty Way Suite 1001 Marina del Rey, California 90292-6695 Internet: smoliar@vaxa.isi.edu "For every human problem, there is a neat, plain solution--and it is always wrong."--H. L. Mencken
geraint@aipna.ed.ac.uk (Geraint Wiggins) (10/19/89)
In article <5013@orca.WV.TEK.COM> steveb@eve.WV.TEK.COM () writes: >I am also looking for information on potential file formats and encoding >schemes for the input side. Any information would be extremely helpful. > >Steve >------------------------------------------------------------------------------- >FROM: STEVEN C. BILOW -- Software Engineer, Tektronix >EMAIL: steveb@orca.WV.TEK.COM PHONE: (503) 685-2463 I and two colleagues in Edinburgh, Alan Smaill and Mitch Harris, have a prototype datastructure based on abstract datatypes which you might be interested in. The advantage is that once you've written your software, anyone can use it on their representation, provided they've followed our (fairly simple) rules. If you're interested, mail me and I'll send you the paper. Geraint -- Geraint A Wiggins | G.A.Wiggins@uk.ac.ed Department of Artificial Intelligence | G.A.Wiggins%ed.ac.uk@nsfnet-relay.ac.uk University of Edinburgh | Opinions are like noses: everyone has 80 South Bridge, Edinburgh, Scotland | his/her own, and most smell...
rwh@xpiinc.UU.NET (Robert W. Holzel) (10/19/89)
This discussion brings to mind the book "Metamagical Themas," which has several sections devoted to whether a computer can be used to analyze creativity. I can't spell the author's name from memory, but he is the author of "Godel, Escher, Bach," and for some time wrote the Metamagical Themas column in "Scientific American."
mgresham@artsnet.UUCP (Mark Gresham) (10/25/89)
In article <10190@venera.isi.edu> smoliar@vaxa.isi.edu.UUCP (Stephen Smoliar) writes: >However, it is unlikely that a machine will ever give you >a "definitive" analysis of a composition, since it is unclear that such an >analysis exists. After all, there're lots of different ways, means and methods for drawing maps (geographical/road/etc.) as well as physical structures (architecture/etc.) so why not music. I think we do tend to forget that intellect (concepts/parsing/etc.) is only a part of human knowledge; it has often undeservedly claimed all of thought as its domain. Well, I won't go into epistomology here, but suffice it to say that I don't think *any* analysis is capable of telling the larger part of the story, although it is necessary for telling its portion. (followup to 'rec.philosophy.amusing' :-) if you like.) Cheers, --Mark ======================================== Mark Gresham ARTSNET Norcross, GA, USA E-mail: ...gatech!artsnet!mgresham or: artsnet!mgresham@gatech.edu ========================================
mgresham@artsnet.UUCP (Mark Gresham) (10/25/89)
In article <267@xpiinc.UU.NET> rwh@xpiinc.UUCP (Robert W. Holzel) writes: >This discussion brings to mind the book "Metamagical Themas," which >has several sections devoted to whether a computer can be used to >analyze creativity. I can't spell the author's name from memory, >but he is the author of "Godel, Escher, Bach," and for some time wrote >the Metamagical Themas column in "Scientific American." And his gross misconceptions about the nature of music are moldy baloney. (Take it from there Steve, you're better on responding about this one than I am!) Cheers, --Mark ======================================== Mark Gresham ARTSNET Norcross, GA, USA E-mail: ...gatech!artsnet!mgresham or: artsnet!mgresham@gatech.edu ========================================
smoliar@vaxa.isi.edu (Stephen Smoliar) (10/26/89)
In article <495@artsnet.UUCP> mgresham@artsnet.UUCP (Mark Gresham) writes: >In article <10190@venera.isi.edu> smoliar@vaxa.isi.edu.UUCP (Stephen Smoliar) writes: >>However, it is unlikely that a machine will ever give you >>a "definitive" analysis of a composition, since it is unclear that such an >>analysis exists. > >After all, there're lots of different ways, means and methods for >drawing maps (geographical/road/etc.) as well as physical >structures (architecture/etc.) so why not music. I think we do >tend to forget that intellect (concepts/parsing/etc.) is only a >part of human knowledge; it has often undeservedly claimed all of >thought as its domain. Well, I won't go into epistomology here, >but suffice it to say that I don't think *any* analysis is capable of >telling the larger part of the story, although it is necessary for >telling its portion. (followup to 'rec.philosophy.amusing' :-) >if you like.) > Have you taken a look at Marvin Minsky's THE SOCIETY OF MIND yet, Mark? Once again, you seem to be heading towards roads he has been trying to build. Actually, you might do better to look at "Music, Mind, and Meaning," which he contributed to the Fall 1981 issue of COMPUTER MUSIC JOURNAL (the same issue which ran my review of GIRDLES, ESSENCE, AND BAGS: THE ETERNAL GILDED SPAGHETTI). THE SOCIETY OF MIND was still a gleam in his eye when Minsky wrote "Music, Mind, and Meaning;" and it is interesting to see how ideas motivated by the study of music led to more general thoughts about cognition. ========================================================================= USPS: Stephen Smoliar USC Information Sciences Institute 4676 Admiralty Way Suite 1001 Marina del Rey, California 90292-6695 Internet: smoliar@vaxa.isi.edu "For every human problem, there is a neat, plain solution--and it is always wrong."--H. L. Mencken
smoliar@vaxa.isi.edu (Stephen Smoliar) (10/26/89)
In article <496@artsnet.UUCP> mgresham@artsnet.UUCP (Mark Gresham) writes: >In article <267@xpiinc.UU.NET> rwh@xpiinc.UUCP (Robert W. Holzel) writes: >>This discussion brings to mind the book "Metamagical Themas," which >>has several sections devoted to whether a computer can be used to >>analyze creativity. I can't spell the author's name from memory, >>but he is the author of "Godel, Escher, Bach," and for some time wrote >>the Metamagical Themas column in "Scientific American." > >And his gross misconceptions about the nature of music are moldy >baloney. (Take it from there Steve, you're better on responding >about this one than I am!) > Robert is probably referring to Doug Hofstadter's essay (as in "attempt"), "Variations on a Theme as the Crux of Creativity." This is less offensive than his book by virtue of being shorter. Nevertheless, it is the usual batch of self-indulgent navel searching. Anyone who gets any of there own ideas by reading it is certainly entitled to them, but my humble opinion is that they would be better off with the Minsky article I cited in my last dispatch. ========================================================================= USPS: Stephen Smoliar USC Information Sciences Institute 4676 Admiralty Way Suite 1001 Marina del Rey, California 90292-6695 Internet: smoliar@vaxa.isi.edu "For every human problem, there is a neat, plain solution--and it is always wrong."--H. L. Mencken
rwh@xpiinc.UU.NET (Robert W. Holzel) (10/26/89)
In article <10274@venera.isi.edu> smoliar@vaxa.isi.edu.UUCP (Stephen Smoliar) writes: >In article <496@artsnet.UUCP> mgresham@artsnet.UUCP (Mark Gresham) writes: >>In article <267@xpiinc.UU.NET> rwh@xpiinc.UUCP (Robert W. Holzel) writes: >>>This discussion brings to mind the book "Metamagical Themas," which >>>has several sections devoted to whether a computer can be used to >>>analyze creativity. >> >>And his gross misconceptions about the nature of music are moldy >>baloney. (Take it from there Steve, you're better on responding >>about this one than I am!) >> >Robert is probably referring to Doug Hofstadter's essay (as in "attempt"), >"Variations on a Theme as the Crux of Creativity." This is less offensive >than his book by virtue of being shorter. Nevertheless, it is the usual >batch of self-indulgent navel searching. Anyone who gets any of there own >ideas by reading it is certainly entitled to them, but my humble opinion is >that they would be better off with the Minsky article I cited in my last >dispatch. Sorry to have dared to tread on matters where you two are apparently the most expert. I'll be sure to keep a close watch on myself in the future. I wouldn't want to threaten anyone's shirt stuffing. Humbly yours, Rob