-- [ From: Richard Tonnesen * EMC.Ver #2.10P ] --

Hello to all...

In keeping with the spirit of this list, I will briefly introduce
myself.  I already know many of you through AMICA, but this is a good
forum to meet others with similar interests.

I joined AMICA in 1972 and bought a Weber Duo-Art grand a few months
later.  I started to look for a good way to apply computer technology
to automatic music, either to preserving the music in machine readable
files, or perhaps to operating a piano in real time from computer
files.  I quickly decided that it would be too much work, too expensive,
and too inconvenient to have a minicomputer attached to a piano, so I
started to consider building a computer-operated roll reader and
companion punch.  It took several years, but I designed and built a 100
channel roll reader based on pneumatic switches, and a 100 channel roll
punch.  Both were operated by a 16 bit minicomputer (one at a time) and
used punched paper tape as an intermediate storage mechanism.  The roll
punch operates at 15 steps per second and punches 45 steps per inch. 
The reader operates at about two feet per minute.  In the reader, the
paper is driven by rubber covered capstans which generate a clock
signal as they turn to provide equally spaced samples regardless of
reading speed.  The system was complete in 1979, and used an Interdata
7/16 minicomputer with 16Kb of core memory that I purchased at a
bankruptcy auction.  Late in the 80's, I converted the main logic of
the reader control program to run on a Macintosh computer and converted
the real-time part (scanning contacts and synchronizing with paper
motion) to run on an Intel 8031 microcomputer.  My main goal was to
take advantage of disk storage instead of paper tape and to use the on-
screen display capability to edit roll images.  As the years went by,
the Interdata machine failed and was beyond economical repair so I had
to convert the punch program to the Mac.  I got a second Mac so I could
run the reader and the punch at the same time.  I had initially written
a primitive editor to run on the Mac, but recently have been using an
editing program written by Richard Brandle of Dallas.  The new program
features conversion of the various reproducing roll data formats to
MIDI output by emulation of the reproducing mechanism.  This equipment
and Brandle's program were used on the recent Gershwin CD by Artis
Wodehouse.

  Recently, I have been looking for ways to make the reading operation
faster and more reliable.  The pneumatic switches need about 5ms to
turn on and 25ms to turn off.  That causes elongation of the note holes
in the copy if paper speed is any faster than about two feet per
minute.  The switches also tend to get out of adjustment.  A few start
to turn on too soon while others start to turn on too late.  They can
be re-adjusted but it is time consuming and exacting work, and really
boooorriiinnnggg.

Roll reading is one of my favorite subjects.  It seems to be very
fashionable lately, but is harder than many seem to realize.  What are
the requirements for a transcription quality reader?  How many samples
per inch is enough? How fast is fast enough? How can you tell when you
have a good file short of punching a new copy and comparing the
original and copy end-to-end, hole-for-hole?

I will be looking for a few good ideas.

R.T.