I think Robert Pintar <pintar@spk.hp.com> has been put up to his
question by Jody, to make us all spill our drinks and choke. Robert
said:

 [ Quoted text deleted by editor, its repeated earlier in this digest

 [ Editor's Note:
 [
 [ I will admit to having an interest in the outcome of this
 [ discussion, as I too have had an interest in writing emulators
 [ for the various reproducing systems.  The reason that _I_ haven't
 [ asked the question previously is that I wasn't ready to deal with
 [ the answers.  The fact that a new subscriber has shown up, with
 [ time on his hands and an interest in this subject, brings me great
 [ pleasure.  No, I didn't put him up to it, but I don't expect anyone
 [ to believe me ;-)
 [
 [ Jody

I am sure plenty of others will communicate the true nature of Duo-Art.
I would like to emphasise the naked difficulties of doing straight
conversions from any reproducing piano system to MIDI.

1. No two reproducing pianos have the same exact tonality.¶
2. There is no standard piano brightness scale on MIDI.¶
3. The apparent powers set by the roll bear an only distant
   relationship to the actual note powers, which depend on
   a large number of obscure factors.

Points 1 and 2 by themselves require any conversion to be a matter of
musical judgement, running auditions on more than one instrument of
each system. You may have to stipulate which MIDI instrument is used to
play the file back and even (if it allows it) what its power settings
are.

Regarding point 3, there was an item in the AMICA Bulletin about two
years ago (I'm not a member so only see it occasionally) about an
Ampico suction meter which wrote a treble/bass suction graph on a
typical roll as it played. Lately Rex Lawson has devised a similar
sensor which writes onto his Apple II roll program the treble/bass
suctions produced in two actual Duo-Art grand pianos by the roll. The
traces for a 1914 (New York) Steinway and 1927 (Paris) G.Gaveau are
different because of markedly different pneumatic sizes and air flows
but have a family resemblance to each other and to the Ampico traces.

Careful examination shows that while the Ampico follows its
instructions more smartly than the Duo-Art, both systems (as one would
expect) are affected by number of notes playing at the time, but also
by rapid passages or trills causing valves to be in transit and
allowing atmosphere direct to the stack and by other pneumatic
operations such as actuation or release of sustaining and soft pedal.
These factors would have been compensated for by the roll editors.

Some literal conversions have been released recently of MIDI to
Duo-Art. One is of Rachmaninov's Prelude Op 23 No 6, by an unknown
pianist, issued by Play-Time rolls (Pagham, West Sussex - see resources
list AMD 96.09.25). An unashamedly romantic piece, this roll "pedals"
(= can be played with foot pedals and hand controls) beautifully, but
the coding looks absurdly flat for a live performance, using the same
accompaniment and theme levels for long stretches, and on a good
Duo-Art piano the performance comes out as simple to the point of
lameness. However, the overall playing level is about right.

So beware ! The intricacies of pneumatic instruments may eventually
with much study be encodable, but the difference between instruments
will surely be the killer factor for a sure-fire sit-back-and-press
-the-key conversion.

Dan Wilson