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

Terry Smythe asked about Bill Flynt's perforator a few days ago.  I had
seen it several times when I first got interested in mechanical music,
and I was thinking about writing a reply, but I decided to see if Bill
would write a description of it himself.  Here is his reply:

          EARLY EFFORTS AT DIGITAL PIANO-ROLL RECORDING
                  by W. E. Flynt, Dallas, Texas

In the early 1970s I had built a crude piano roll perforating machine,
in an effort to duplicate antique paper rolls for my Ampico reproducing
piano collection.  As an adjunct to that purpose, it occurred to me
that there might be a possibility of making "live" recordings, as was
done in the hey-day of the player-piano (1920s).

However, even if I were to have equipped my piano keyboard with under-
the-key electrical contacts, the perforating machine could not possibly
do its perforating job in real time.

My task, then, was to slow down the data rate enough to allow the
perforator to respond to the key-contact closures, as it punched holes
in the blank paper roll.

This was the era of digital tape recording (before personal computers,
magnetic disks, etc., etc.), and there were many magazine articles
dealing with digital encoding, clocking, etc., using early TTL chip
technology.  My format allowed the scanning of 128 separate channels of
binary information, although only 80 channels were needed for the piano
rolls.  Frame rate was 100 per second.

A popular data transmission method at that time was the so-called
Manchester code, in which the sequential digital data was synchronized
with its own clock rate in a single data stream which could be recorded
on, for example, -inch magnetic tape.  I found that this format was
amenable to commonly available audio tape decks at that time which
would operate at several tape velocities, selected mechanically by the
operator, via pulleys, capstans, etc.  The recorded piano performance
(under-key contact closures) could be "played back" and decoded, for
example, at 1/8th of the recorded speed, thus allowing the perforating
machine to follow the data easily...  although requiring 8 times the
performance duration to complete the task.  Recording was at 15 inches
per second, and playback was at 1.875 ips.

This was a very crude scheme, requiring much hand-editing, but I was
able to demonstrate it at the 1973 national convention of the Automatic
Musical Instrument Collectors' Association (AMICA) in Ft. Worth.  About
6 to 8 "live recording" piano rolls were made (principally in the
Ampico reproducing roll format) and sold to the collector trade prior
to a workshop fire which destroyed the perforating machine in 1976.

Nowdays, of course, there are far more elegant ways to record piano
performances digitally (MIDI, et al), but to my knowledge, this was the
first effort of its kind.

5-22-96