Marc,
   I'm always interested in new methods of conversion.  Do you know what
the "straightforward manner" is that is used by  Mr. Heyworth ?

Here's a quick (and overly terse) review of what I remember about "roll
digitizing projects":

One of the advantages of optical scanning is that you can resolve a spot
much smaller than a tracker bar hole.  You save the image data and then
apply whatever "tracker bar algorithm" suits you later.  Remember that
on some instruments the tracker bar holes are not identical (some are not
the same shape and some come slightly early).  The properly adjusted pneumatic
reader with micro-switches on the pouches will do a great job on a roll
intended for the particular tracker bar, but the "interpretation" is done
pneumatically [Wayne Sankhe and others use this method.  For good quality
control, there is quite a bit of monitoring/adjusting that goes into the
process.  Its not a "one size fits all" solution.  Although Mills Novelty
company used wire brushes through the paper to read rolls, (and David
Wasson has built a Midi reader for his hand-cut band organ rolls this
way) I'm skeptical that wire brushes are a good idea, because of noise
(contact bounce) and wear-and-tear on the roll.  One of our readers has
built a "scanner" by stuffing photo-diodes into the holes of a tracker
bar.  This system has some problems with translucent paper and doesn't
exactly reproduce the behavior of the pneumatic pouches, but is very
clever, inexpensive solution.  Applying polarizing filters to the system
might solve the translucent paper problem.  The direct optical scanning
approach also has its problems, with ink and other artifacts on/in the
paper (word rolls must be scanned from the back).  A full-color scanner
might provide for correct interpretation in spite of paper artifacts by
some clever multi-color lighting scheme.  Kravitz/Ames and Zoltan Janosy
both have experimented with monochrome optical scanning with good results.
Both of those projects had their problems.  Zoltan contracted out the
scanning, which was both slow and expensive (but very accurate) and
Kravitz/Ames used an inexpensive fax scanner (Chinon DS-3000 with modified
firmware) but it is slow (40 scans/second), and is not grey-scale.
I (Kravitz) have recently purchased a Hewlett-Packard Scanjet 3p, which
is 8-bit monochrome (grey-scale) and may be able to scan as fast as
120 lines/second.  If its firmware could be modified to scan continuously,
it might represent the cost-breakthrough which is needed.  I paid $315.00
plus  $18.00 for shipping to USA-Flex (mail-order).  I did the reverse-
engineering on the Chinon Firmware.  I'm not sure I want to do another one!
I'd be happy to consult/advise anyone who wants to embark on that project.
A guy at the MIT medial lab has done two scanning projects, one using a
linear CCD array (similar to a fax scanner) and then later using a standard
NTSC camcorder and a "frame grabber".  He felt that this method allowed
him to get the temporal information more easily (all the other methods
require some kind of paper speedometer as an auxiliary input the the
recording computer).  I'm personally skeptical that the frame grabber
solution works for all rolls (we've been scanning Aeolean Residence Organ
rolls which are around 178 holes wide) because of the limited horizontal
resolution, but prices for NTSC (standard US TV) cameras and frame
grabbers are dropping dramatically.  Putting the frames back together can
be "interesting.  When Mike and I started our project we were using a
386/SX-25.  Now P5-90's are available for same price and P5-133's or better
are available if necessary.  There's still the issue of paper-transport,
which is non-trivial if you want to be sure an old roll doesn't get "eaten".

Please don't give up in your pursuit of digitized roll music!

Jody

P.S.  I'd really like to hear from others who have done or thought about
roll digitizing.  There's an amazing amount of material out there to
be digitized...