Janosy and Macsai seek Welte-Mignon player acceleration/deceleration
data for their project of reproducing the music from the transcribed
rolls,  using a Welte-to-Midi emulator and a synthsizer or
Disklavier.  Virtually all classical rolls include inherent time
correction which attempts to compensate for the increasing paper
velocity as the diameter increases on the take-up spool.  But playing
the transcribed roll via Midi fails to duplicate the action of the
spoolbox, and so "the music slows down." I can't think of any
reliable method to obtain the data they need by direct measurement of
the paper speed:  every airmotor system has too many uncontrolled
variables.

Rick Pargeter, in digest 950614, said that he must continually move
the Tempo lever as he plays a long roll on his piano, in order to
keep the music speed constant.  I have the same problem with my
airmotor piano, and I do the same "manual deceleration" during
playback.  I am aware that the piano would do better if there were no
leaks and the bearings were perfect, et cetera.  It has been 5 years
since the airmotor and governor were re-covered, and I don't know how
long since new felt was fitted in the bearings.  The piano is almost
80 years old, and just "typical", I suppose.

My Ampico B piano with governed electric drive motor also requires
"manual deceleration", but not as much as the airmotor piano with the
smaller diameter spool.  Not all rolls demand playback speed
adjustment, but most of my favorite pop tunes *do*.  It seems, with
these songs, that the mind is as sensitive to tempo change as it is
to pitch change: one percent is generally discernable.  (And everyone
knows the seasickness that accompanies the pitch change as a
phonograph spring-motor slows down, or the battery in a cassette
player dies!)

I have adequate data from Ampico roll analysis and literature which
suggests that a take-up spool of 4 or 5 inches diameter, as I recall,
would provide the proper acceleration of the roll for a spool turning
at constant revolutions per minute.  This data assumes that the
average thickness of the paper on the spool is about .003 inch (.075
mm).  From this, we could derive the optimum airmotor deceleration
(only for Ampico) due to the constantly increasing torque.  (I will
have to find my notes about this for you, Zoli.)

Notice that I say "optimum".  The speed-versus-torque characteristic
of the airmotor and companion pressure regulator ("governor") are
very dependent upon system friction, and upon air leaking into the
airmotor and governor pneumatic.  I don't know of any standards for
airmotor systems regarding the recommended friction of the airmotor
bearings, and there is no way to compensate for air leaks while
retaining the same speed-torque characteristics.  There will be very
big differences between pianos.

It seems to me that we must rely upon our ears and good musical
judgement to re-create the proper "tempi" throughout as the music is
played.

-- Robbie Rhodes,  15 July 1995