Jody said he'd like to see some more technical topics kicked about in
this interest group, so here are some opinions of mine which should
generate discussion.  As you can tell, I'm interested in the
techniques and controls (testing, etc.) required for the production
of high-quality music rolls.

** "Who can copy my (reproducing) piano roll?"

These folks are active, to varying degrees, in roll copying:

1.  Direct asynchronous copying without computer files:

   Keystone Music Rolls has the Larry Givens high-res perf, approx 60
steps/inch.  I believe this machine is used for all Keystone
production now.

   John Malone, PlayRite, built his own high-speed perf which can run
at 3000 rpm, approx 60 steps/inch.

   QRS (capabilities not known)
   Mike Kittner (book music and rolls) 
   Don Rand (6 ch/in nickleodeon rolls) 
   L. Douglas Henderson (Leabarjan frame punch)

2. Asynchronous copying using intermediate computer file:

   Carl Lambie - reader & perf, Midi controlled

   Richard Tonnesen, Custom Music Rolls - reader & perf, unique file
   format, Midi file conversion available, 45 steps/inch

   Siegfried Wendel, Mechanical Musicboxes Manufactory GmbH,
   Ruedesheim, Germany - single-hole frame punch with servo
   positioning, experimenting with optical reader.

4. Reading only:

   Mike Ames/Jody Kravitz - optical line scanner
   Wayne Stahnke - pneumatic reader, 0.01 mm/frame.

** "Is any company using master rolls and synchronous perforating?"

   There is no one I know of who is routinely perforating production
rolls using 2-to-1 or 3-to-1 master rolls, or the computer
equivalent, as in the 1920's.

   Keystone Music Rolls in Pennsylvania has the capability of
producing rolls from their collection of Ampico production masters,
but the old master rolls are far too fragile today, and so all of
Keystone's production is 1-to-1 asynchronous copying of orignals
using primarily the Larry Givens perforator assembled many years ago.
This is the same machine that was purchased by Harold Powell and used
at Klavier for copying at about 60 steps/inch.  Duo-Art copying
capability (the snakebites) was added by Mike Ames.  Keystone may
also be using the "Pianola" perforators built in Los Angeles in the
60's.

   QRS, to my knowledge, has transcribed all their "masters" to Apple
2E computer files and has small computers controlling the perforators
now.  From the looks of their current roll production, and a couple
of disk files I've seen, they may not be using a truly synchronous
system.

   PlayRite does all production as 1-to-1 asynchronous copying.

   Richard Tonnesen, Custom Music Rolls, has a computer-controlled
synchronous perforator which was used to make the production rolls
for Chicago AMICA of the Leo Podolsky Ampico performances.  Since the
original recordings (made on a Stahnke SE piano) were edited for
production at 45 steps per inch (Tonnesen's perforator), this project
qualifies, in a sense, as a Synchronous process.  Certainly all the
production copies (about 70 sets, I believe) can be found to match
hole-for-hole, just as 1920's Ampico production.  But, other than a
few special projects like this, the bulk of Tonnesen's output is from
asynchronous reading of original rolls, perforated synchronously.

** "How do you define synchronous production?"

   Well, ...  We're obviously talking about quantized binary data,
since that's what a piano roll is. The simple definition is that all
production copies, from all production batches, will match
hole-for-hole, just as today's compact discs (CD's) for music or
computer data perfectly replicate every data bit.

** "Why do you need a synchronous process?"

   In order to maintain the timing relationships to precise
tolerances.  Consider the two production schemes used by Welte:  
all of the Welte rolls perforated in America were synchronous, from
sprocket-drive master rolls, but the Red rolls perforated in Freiburg
were asynchronous.  Why the change?  Most likely because they
realized that they needed better control of the hole placement, or
the Welte "sforzandos" would be improperly timed.  So when Edwin
Welte set up the shop in Poughkeepsie he used the American
synchronous perforating process.

   Much the same reasoning is behind digital versus analog sound
recording, and for the same reason.  The early phonograph sound
recordings were produced by analog replication of the disc medium --
they stamped a wax or shellac copy of the disc from a negative mold.
Magnetic tape is simply copied from a master tape to an analog copy,
and it similarly suffers from noise.  But the ear is pretty good at
ignoring noise, and so both processes were hugely successful.

   With digital audio recording, like CDs and DAT, the analog data is
sampled at a high frequency, the signal samples are converted to
binary (on-off) bits, and the binary data is replicated by a
controlled process which assures no corruption.

   Prodution copies of contemporary CD's (either of music or CD-ROM
data discs) are supposed to match, bit-for-bit, across the whole
disc.  Production copies of American rolls of the 1920's match hole
for hole, unless the perforator had sluggish channels (like due to
clogged bleeds).

   One may argue that a high-resolution asynchronous copier, such as
at PlayRite and Keystone, should be adequate.  But simply "adequate"
is not a good qualifier. One might better ask:

** "Will a recut of an Ampico roll, orignally perforated at 30 steps
    per inch, and re-sampled at 60 steps per inch, sound as good
    as the original?"

   No.  Nor will it if re-sampled at even 100 steps/inch.  This was
proven in a series of careful "comparison listening" tests performed
at Dick Carty's piano shop by a *very* critical panel: the crowd of
Los Angeles region player and reproducing piano technicians!  In the
early 70's Wayne Stahnke was developing the IMI Cassette Converter,
which substituted magnetic tape cassettes for the paper roll of an
Ampico, and he wanted to determine the slowest sampling rate needed.
The listening panel and he judged 100 samples per second as
"adequate", and that became the (scanning) frame rate of the Cassette
Converter. But close comparision with the original roll on a really
good Ampico A piano still displays missing accents because the
intensity commands fail to occur ahead of the note or chord. ... 
One should best ask:

** "How do you assess, quantitatively, the precision of a music roll
copying and production process?"

   Ah, good question.  We will discuss this later.  

   For now, why don't you find an old original and a new copy of the
same performance, and lay them together on the kitchen table.  If
they *really* match, hole-for-hole, your original may be counterfeit.

   Think about that...!

-- Robbie Rhodes, 1 July 1995