Robbie, you wrote [in a comment in MMD 981023]:

 [ No computer programs exist yet which can convert a MIDI file
 [ with MIDI "velocity" (such as recorded by a Disklavier) into
 [ a music roll for a pneumatic reproducing piano.  All the codes
 [ for the pneumatic expression mechanism must be entered manually,
 [ just as the editors did long ago.  But we're slowly getting
 [ closer to this goal !
 [
 [ -- Robbie

Robbie, could you please elaborate on the progress of MIDI "velocity"
to Ampico Expression.  Didn't you do this to some extent when you
worked on the Podolsky Ampico rolls? What are the problems that inhibit
progress towards a computer program that will produce good Ampico
expression?  How far away is this goal?

It seems to me that a program like this could be the so-called "Holy
Grail" of roll conversion.  It would finally make possible the re-coding
of one expression format ( Duo-Art or Welte) to Ampico.  This could be
accomplished in several ways.  MMD member Tom Steuer has a Vorsetzer
that will play  Ampico, Duo-Art and Welte rolls hooked up to his
Disklavier.  He can record all of these rolls as MIDI output I believe.

If we could merge the converted MIDI velocity coding with an accu-
rate note scan of the rolls, Ampico owners could have many great
performances available that were found previously only on other
reproducing rolls.

Another way would be to create a program to turn Duo-Art or Welte
expression into MIDI velocities.  This could create many new possibili-
ties for the PowerRoll too.  Of course this will make it much easier for
all the talented pianists and arrangers out there to make new Ampico
rolls.  This one piece of software could be a major advance for the
reproducing piano.

Is this a pipe dream or will it be a possibility someday?

Michael Swanson

 [ Programmer's reply:
 [
 [ I tinkered with it briefly and quickly realized it would take
 [ months of full-time work to write a computer program, so instead
 [ Bob Taylor (now of Columbia, MO) added the Ampico expression codes
 [ manually, guided by a list I provided of the hammer velocity of every
 [ note.  It took Bob many weeks of work to edit and add expression to
 [ almost 50 minutes of music!  The final product was very well received
 [ by the subscribers to the project; I believe that about 70 sets of
 [ rolls were sold.
 [
 [ Based upon my brief experiments I believe that an expression editing
 [ program with satisfactory performance can be created with about the
 [ same amount of effort (man-hours) as Richard Brandle has put into
 [ the development of "Wind".  (How much is that, Richard?  ;)   Even
 [ with the editing program, though, I estimate that 10 minutes of
 [ finished music will require about 50 minutes of editing time, which
 [ is a ratio of 5 to 1.  That's a marked improvement over Ampico in
 [ the 20s, when a typical classical song took an editor perhaps three
 [ days.  The editing norm at Marantz Pianocorder was 30 seconds of
 [ finished music in 8 man-hours labor !
 [
 [ An expression editor relies on a built-in piano emulator program for
 [ its coding accuracy; several folks are working now on that task.
 [ That's about the only knotty problem left.  (An emulator is a
 [ program or sub-routine which accepts a music roll input and creates
 [ the resulting hammer velocities by modeling the real-life pneumatic
 [ piano system.)  With emulators for Ampico, Welte, Duo-Art and Hupfeld
 [ it would be fairly easy to convert one roll format into another.
 [
 [ When a market develops which will pay for the program development
 [ I'm sure someone will write the program, and new rolls (or conver-
 [ sions) will be available for reproducing pianos.
 [
 [ -- Robbie