It's great to read all the explanation forwarded by all these
knowledgeable people.  I would like to get some more explanation about
the part where Bob Baker mentions the delay that a solenoid introduces.
You must know that the drum [actuator] design is very different from
a piano key.

When we first started the design of the velocity drum set, we expected
to get a lot of timing problems that we had to cover by difficult
software.  To our surprise, there wasn't any timing problem!

Simply explained, the way we it works is that the stick is always on
the instrument, and gets pulled away by a solenoid.  Then a spring
returns the stick to the surface.  At it lowest velocity, you almost
don't see the stick moving, but still you hear the sound.  At its
highest velocity, the stick moves about 12 centimeters, and a loud
sound is produced.

The stick returns always exactly the same time after the start of the
note, and there is not even one millisecond error over the whole
velocity scale.  So the solenoid and spring are behaving perfectly.

A piano is very different.  The hammer must always travel the same
distance, and a loud note is produced by moving the hammer faster,
a soft note by moving the hammer slower.  This means that you must
start to play a soft note earlier then a loud note, if you want the
notes to arrive at the same time.

So, if you want to record a piano performance, you must detect the
moment a key starts moving and the time it takes to hit the bottom,
and store this information.  To reproduce this performance, you must
send the soft notes sooner then the loud notes if again you want both
notes to arrive at the same time.  This goes for any system, let it be
air, vacuum or solenoid.  You can do it by recording the exact time
when the key starts moving, but this way you could not play that file
on a module, as the soft notes would play to soon.

An other way would be to record the moment the key is down, store that
exact time along with the time it took to come down, and store that as
the velocity value.  Then you could change the start time according the
velocity.  Electronically, this can be done by a processor.

My question is: Am I correct with my assumptions regarding the piano,
and if so, how is that handled with a paper roll?  Are soft notes moved
forward on the paper roll?  If so, you still have a problem if you want
to change the overall volume.

Remember that we have to get things to play together.  So if it is no
problem if a solo piano is moving in time when you change it's master
volume, it becomes one the moment you have an accordion or drum set or
sound module play together with the piano.

Tony Decap
Tony & Frank Decap, DECAP Herentals, Belgium
http://www.decap-herentals.be/

 [ You're correct: the positions of the soft notes on the music roll
 [ typically are advanced slightly from the nearby loud notes.  The
 [ timing error introduced by altering the piano volume setting (e.g.,
 [ "Subdued" or Brilliant") isn't noticeable in a properly maintained
 [ piano and so it's not considered a problem.
 [
 [ When Yamaha and QRS instruments are connected to an external
 [ digital controller, the piano hammer is supposed to strike the
 [ string exactly one-half second (500 ms.) after the command is
 [ received via the MIDI wireline.
 [
 [ In music rolls for Wurlitzer band organs, the bass drum hole is
 [ advanced 0.050" to 0.100" ahead of the notes, which presumably
 [ corrects the error of a bass drum beater with 90 to 180 ms. delay.
 [ The compensation observed on Wurlitzer rolls that I've seen isn't
 [ consistent, and many band organ owners don't know how to adjust the
 [ beater travel distance and the valves and springs to eliminate the
 [ timing error.
 [
 [  -- Robbie