To continue the recent discussion about scanning music rolls:

I found an interesting component in this month's DigiKey catalog
(a mail-order electronic parts supplier).  They have a visible laser diode
line generator (I assume this is the kind designed for scanning barcodes
at checkout counters) for $85.00, manufactured by Applied Laser Systems.
This tiny little device will project a 1mm wide LINE at an 85 degree
beam angle, at a wavelength of 670nm and output power of 2.5mW.
Positioned just 6 inches above a page, it would create a line 11 inches
long and approximately 1mm wide.

It occurred to me that this might be a good exciter lamp for scanning
music rolls-- the narrow line would help ensure that the same point
in time is being read across the width of the roll.  Rather than using
two offset fax scanner pickups directly against the paper, how about
using one pickup and optically compressing the laser line onto it
somewhere beneath the tracker bar?  (The tracker bar would be modified
to have one narrow open slit all the way across instead of the usual
holes).  Rough diagram:

•  laser line generator
           |
          / \
        /     \   line projected down onto music roll and through tracker bar
      /         \
    /             \
_______________________ music roll
======================= tracker bar
      \         /

          < > lens
         /   \
        ------- fax scanner head

•Of course, this is assuming that the one fax scanner has sufficient
resolution for what we're trying to do.  I'm skeptical that we could
get these scanner heads to pump data fast enough to scan rolls at
full speed.  Has anyone investigated this issue yet?

In one of the recent digests, it was suggested that we should be able
to scan 12 rolls an hour, if I remember correctly.  No offense, but I
think this is not at all realistic.

I'm involved in a project to digitally convert over 100 Pianocorder
tapes into MIDI format, producing a separate MIDI file for each
song/medley.  The title/composer/pianist/catalog #/etc are all embedded
in the MIDI files as text events.  Even with semi-automatic software
performing the music conversions, it takes a certain amount of time
for me to key in the text information, apply it to the correct songs,
listen to the files to make sure they're okay, organize the data
in appropriate directories on disk, choose appropriate filenames,
check the log files generated by the conversion software...
There's a lot of little tasks involved, and it takes me at least
an hour to scan and archive just one tape (about 40 minutes of music).

Remember, I'm just reading perfect digital data off the tapes-
I'm getting the music accurately every time.  Scanning real piano
rolls will present a whole slew of problems... from tears, repairs,
sticky tape, alignment errors... almost every roll will require some
editing.  What do you do about rolls that are missing their labels?
You can't just do "noidea1.mid, noidea2.mid" and move on.  Researching
missing information like this will take time too.

The idea of someone jetting around the country with a laptop computer
and portable roll scanner (that handles every known format), gradually
accumulating an authoritative archive of rolls for posterity--- it's a
nice idea, but there's a lot of work beyond the actual scanning itself,
and that's after we get a system designed and built.  We really ought
to look into what funding might be available for this project.