Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!yale!umich!sharkey!math.lsa.umich.edu!math.lsa.umich.edu!hyc
From: hyc@math.lsa.umich.edu (Howard Chu)
Newsgroups: comp.sys.next
Subject: Re: Sheet Music Editor
Message-ID: <1990Jun2.050130.2503@math.lsa.umich.edu>
Date: 2 Jun 90 05:01:30 GMT
References: <1990May9.051540.4355@dept.csci.unt.edu> <15282@s.ms.uky.edu> <1990May21.153705.25726@dept.csci.unt.edu>
Sender: usenet@math.lsa.umich.edu
Distribution: usa
Organization: University of Michigan Math Dept., Ann Arbor
Lines: 46

In article <1990May21.153705.25726@dept.csci.unt.edu> doug@dept.csci.unt.edu (Douglas Scott) writes:
>In article <15282@s.ms.uky.edu> libacct@ms.uky.edu (John Coppinger) writes:
>>I think there's a gold mine to be claimed by the person or people who can
>>come up with a system that can scan sheet music and convert the bitmap
>>into a flawlessly reproduced music representation file, such as a ScoreFile.
>
Glad to see I'm not the only one in the world thinking about this. I
don't particularly care if I can get a flawless score out of it, but it
would be a lot more convenient for getting large amounts of music into
electronic format, for, say, MIDI playback.
>[much stuff deleted]
>
>>     Building the library doesn't seem trivial, though.  The variety
>>of musical notations would make it a computer vision expert's 
>>nightmare, or dream, depending on his attitude, I guess.  Maybe, with
>>the help of massively parallel systems like Danny Hillis' Connection
>>Machine, the problem would become a simple one.  
>
>Think how long it took to develop the technology to just read in typed
>characters!  The trouble I see with just "reading in" musical notation is
>the degree of complexity in even a straightforward, standard score.  What
>about slurs that run across two pages, for example?  Anyway, I am sure that
>at some point anything will be possible, but I  shudder to think of the first
>20 versions of the software...given how many mistakes a professional music
>copiest (who supposedly understands the context) makes, what will the program
>be like?  Well, at least it wont be ME doing the R & D.

Just because it'll take maybe 20 or more revisions to get it "perfect"
doesn't mean it's not worth doing, eh? (I mean heck, look at Emacs! Er,
on second thought...  }-) Recognizing typed characters seems to be a
pretty reasonably solved problem now. At the very least, you should
be able to recognize printed musical scores. And even if the first pass
isn't perfect, if you mate this with a decent scoring/editing system, it'll
be a lot faster to use than for a poor human to enter an entire score by
hand. The music industry seems to be doing a good business selling prerecorded
tunes on 3.5" floppy disks. Maybe it won't be a gold mine, but I think
there'd be a pretty substantial market for music scanning software... And
again, it's no good if all you can do is scan it - you need to be able to
manipulate the score once it's in memory. I'd like something along the lines
of Audio-Light's Music Studio program, or C-Lab's Notator. Something that
lets you assign parts to MIDI instruments/channels, etc. It's pretty tedious
using these programs to enter music by hand... This sort of thing would
be a godsend for electronic musicians...
--
  -- Howard Chu @ University of Michigan
  ... the glass is always greener on the side ...