Mr. Ryan's posting yesterday, and Mr. Broadmoore's clarifying
remarks posted this morning, fascinate me.  Even though I have an
Ampico, there are many situations when I would still like to 'play'
the pieces through my computer's MIDI system.  I would place my order
today, except that there is still a big 'hole-in-the bucket' in my
understanding of how the combination of the PowerRoll CD and WindPlay
software is supposed to work.  Perhaps you can clarify further.

When Mr. Stahnke's Ampico CD is used to actuate the PowerRoll,
all it does -- in spite of all the sophistication underlying the
CD-ROMs information content and the PC logic that is used with it --
is to 'move' the pattern of control holes in the original roll,
to commands which open the tracker bar ports on the user's piano
the same way that the roll does.  It remains the responsibility of
the PIANO to convert the pattern of control port openings and
closings to the dynamic settings needed to produce a realistic,
expressive performance.

MIDI players, no matter how sophisticated, have no power to do this.
Somewhere, there must be inserted into the chain of events, some sort
of 'model' which interprets and converts the roll coding instructions
into a series of 'velocity instructions' for each note, which the MIDI
player uses to set the dynamics of the electronic performance.

Therefore, two questions:

1)  Where does this model reside?  There are two possibilities:
Either Mr. Stahnke's CD does contain 'velocity instructions' on the
individual notes, even though these are unneeded and ignored when the
CD is used with a PowerRoll, or the WindPlay contains the 'Ampico
model' which executes the code interpretation in real-time, as the CD's
information is translated into MIDI instructions.

2)  What is the nature of this model (whether located in Mr. Stahnke's
studio or in the WindPlay software) and how does it work?  It is easy
to construct a 'static' model of Ampico response.  (See AMICA Bulletin
Vol. 34, No 4, D. Saul, "Performance Characterization...Ampico A..,")

    But I know enough about fluid control to know that such a model is
only the tip of the iceberg.  Effects of regulator load, valve seat
forces and pressure drop, and other short-term phenomena too complex to
go into here, make a truly rigorous model of the Ampico system response
either difficult or impossible to realize.

    I know that Mr. Stahnke claims to be able to do this.  I have his
Rachmaninoff  CD, and that proves it is possible.  But in that instance
he had the advantage of plenty of time to make the 'conversion', and
the opportunity to edit the results.  I remain skeptical of the claim
that it is possible to rigorously model the Ampico in real-time.

If either of you gentlemen can clear up these points for me, my check
will be in the mail.

Richard Vance


 [ Horst Mohr's "Electronic Ampico" system includes an optical roll
 [ reader to play an Ampico roll.  Horst wrote a computer program --
 [ an 'emulator' -- which process the signals from the tracker-bar
 [ photocells into hammer-velocity commands, expressed as MIDI wireline
 [ signals to a piano synthesizer.  The MIDI output plays in real-time,
 [ just the same as the pneumatic system, and with the same delays from
 [ the "note-on" valve command until the sound is heard.  It's simple,
 [ because all of the time-dependent parameters (e.g., the fast and slow
 [ crescendo commands) are already given _on the roll_ with plenty of
 [ 'lead time'.  Richard Brandle's program includes similar emulators,
 [ but reads input data from a disk file instead of a tracker-bar.
 [
 [ The performance of the emulator depends upon how detailed is the
 [ model of the pneumatic system, as well as the accuracy of the
 [ model's parameters.
 [
 [ When used in conjunction with the PowerRoll and a real Ampico piano,
 [ the "WindPlay" program does not use it's built-in Ampico emulator;
 [ it simply converts the perforator control data from Wayne Stahnke's
 [ CDROM into commands which turn on the electric valves in the
 [ PowerRoll unit.  The Ampico pneumatic reproducer does the rest.
 [ MIDI wireline signals are used as a convenient means to connect
 [ the PC to the PowerRoll; the MIDI 'velocity' byte in the wireline
 [ signal is constant, however, since it only turns a valve on and off.
 [
 [ -- Robbie