daemon@BARTOK.ENG.SUN.COM (09/17/89)
Music-Research Digest Sun, 17 Sep 89 Volume 4 : Issue 56
Today's Topics:
Administrivia: sleepy Digest for 3 weeks
Machine-readable Bach (2 msgs)
MTEX User's Guide
*** Send contributions to Music-Research@uk.ac.oxford.prg
*** Send administrative requests to Music-Research-Request
*** Overseas users should reverse UK addresses and give gateway if necessary
*** e.g. Music-Research@prg.oxford.ac.uk
*** or Music-Research%prg.oxford.ac.uk@nsfnet-relay.ac.uk
----------------------------------------------------------------------
Date: Sun, 17 Sep 89 08:55:30 bst
From: Stephen Page <sdpage@uk.ac.oxford.prg>
Subject: Administrivia: sleepy Digest for 3 weeks
To: music-research
There won't be a Digest for the next three weeks, as I'm on honeymoon
in various parts of the world.
------------------------------
Date: Sat, 9 Sep 89 20:50:20 MDT
From: ames!cs.utexas.edu!watmath!calgary!conklin@uk.ac.oxford.prg>
Subject: Machine-readable Bach
To: Sun.COM!bradr@cs.utexas.edu
In response to Dr. Page's comment on machine-readable Bach
in MR digest.4.55. I am not familiar with your
thesis...do you have an abstract on-line you could send
to the digest? Also, I think it would be nice if more
contributors perhaps gave a short description of their
own work and backgrounds!
Oh, by the way, let me introduce myself. I'm doing an
MSc in Comp. Sci., researching prediction, compression,
and modelling of music. I am finding the current flurry
of transactions on Markov models of music somewhat
disappointing - in particular with respect to the lack of
references cited and theoretical discussion. I have an
extensive bibliography on the subject that I will send
to the digest, if it would be of interest.
I have developed some techniques for handling the
multi-leveled strata present in music, and constructing
probabilistic models of these strata, which are later
used in compression. The models are multi-tiered as well
as linear. This means that context for prediction is
not only 'n' preceeding events: the biggest problem with
applying standard Markov models to music. Context can
be based on preceding events *on some level*, or events
*on higher levels*. The structured representation of
a melody is induced by the application of a set of
context-sensitive conditional rewrite rules. I am
exploring whether such models have more predictive power
than standard, multi-viewpoint Markov models.
Being musicians, we hypothesize, of course, that such
models will be a vast improvement over strict linear
models, provided that the rewrite rules are constructed
with knowledge and care.
I had originally intended on working with a subset
of Bach fugues. In fact, I even ordered the WTC from
Oxford, mentioned by sdpage. I found out (very quickly!)
that the Fugues were much too complex a point to start
from - so I decided to concentrate on the single melodic
line of Bach Chorales.
Not that I've put a lot of effort into it, but what
do the block of codes after the BEGIN mean? Obviously
keysig, timesig etc., but let's see a formal description
of the language! If you've completely decoded this
stuff (as it sounds like Dr. Page has), I'd be very
interested in hearing your exact interpretation.
By the way, I have close to 100 Chorales encoded
in my own little representation language (not unlike
Hewlett's). If you're interested in seeing them and/or
the representation language let me know.
darrell conklin
conklin@cpsc.ucalgary.ca
[ The abstract of my thesis was included in the Digest, volume 3 issue
36. You can get a back issue from the Archive Server. Anyone wanting
a copy of the thesis itself should write to the Bodleian Library,
Oxford.
I have hunted several times for my instructions on the BEGIN block
of the CCARH Bach texts - I promised this to Lou Burnard of the
Oxford Text Archive years ago and haven't forgotten! I'll have another
look, but if it doesn't turn up perhaps I can persuade CCARH to
send a note. I think it tells about the edition, the work BWV number and
movement ID, and the number of voices (track ID info). - SP ]
------------------------------
Date: Sun, 10 Sep 89 00:40:00 EDT
From: "Sterling Beckwith (York University)" <GUEST4%bitnet.yusol@ca.utoronto.utcs.ugw>
Subject: Machine-readable Bach
To: music-research@com.sun.eng.bartok
I'm afraid Usenet seems the more appropriate forum for Mr. Ioannidis' request
after all, though our moderator should be congratulated for turning it into
a useful and informative item for the Digest.
You don't write to a Research List asking for "some Bach in machine-readable
form", without specifying what form, what machine, and what uses you have
in mind -- not in a field which is still struggling to arrive at standards
for so complex a range of purposes and/or needs.
It's always nice, though, to see someone get better than they deserve.
S. Beckwith, York University, Toronto
------------------------------
Date: 9 Sep 89 11:08:11 GMT
From: Francois Jalbert <ubc-cs!grads.cs.ubc.ca!jalbert@edu.washington.cs.beaver>
Subject: MTEX User's Guide
To: music-research@uk.ac.oxford.prg
There has been some talk in the last few months about the music
typesetting set of macros MTeX. Since I came across that package last
April, I have looked somewhat into it. I would like to share my findings
with all of you people out there.
MTeX can handle single-staff music with one or two voices. It can also
put text below the staff, and knows about most accents that go above
notes, or simply above the staff. However, it cannot deal with multiple-
staffs, or with notes of variable size. The current version is only 0.92
and one can wonder if a version 1.0 ever saw the light of day. I am
still hoping one day to finally manage to contact the authors. So far,
no luck in my attemps.
I immediately saw last April the potential of this system which happened
to answer exactly my modest needs. Unfortunately, the brief document in
german enumerating the available macros was the only documentation
available. I felt the need for a more complete user's guide still
remained unanswered, and so I volunteered to give it a try.
The guide (Version 1.1) has been around for some time and is about 40
pages long. Among other things of interest, Appendix C lists several
errors I already have found in MTeX. I also generated several standard
series of fonts which accompany the guide. The whole MuTeX package
(MTeX+Guide+Fonts) can be FTPed from the following three locations
stolaf.edu 130.71.128.1 pub/MuTeX.tar.Z
yale-zoo-suned.arpa (suned.zoo.cs.yale.edu) 128.36.21.1
cs.ubc.ca 128.189.97.5 src/MuTeX/MuTeX.tar.Z (slow)
If you plan on making this package available for FTP at your site, I
would appreciate you letting me know your coordinates and the path to be
used for retrieving MuTeX. I plan on keeping updating the list of these
sites in the Preface of the forthcoming versions of this guide.
Francois Jalbert
'
jalbert@cs.ubc.ca
------------------------------
End of Music-Research Digest