The process I use to remove old pneumatic fabric I share with caution.
I've been doing this for some 30 years and it works for me, but it can
be dangerous, not only for the bellows being worked on, but also on
self and home.  It involves heat, lots of it, but with a measure of
discretion and precision.

I agree with steam/hot water for leather removal, but in small
amounts.  Very easy to overdo it, and possibly warp the lumber or
separate the plys.  Use water sparingly.

For small pneumatics, I put them into a conventional sandwich toaster
that I picked up at a garage sale for $2.  It has a thermostat which I
can set just a little above lowest possible temperature.  I fill it up
with about 30 note pneumatics, let soak in the heat for about 20
minutes, then go at it.  As I remove one pneumatic, I put in another on
the LIFO principle.

Both the fabric and the glue are well softened, and the fabric removal
is clean and simple.  Entire stack of 80+ pneumatics can be stripped in
about 30-45 minutes.  The residual animal hide glue is very slight and
can be lightly sanded off real quick on a piece of medium sandpaper
affixed to some solid scrap lumber.  That step might take another 45
minutes.  I finish up with a fine file removing all the sharp edges off
every pneumatic.  I do not want to risk having new fabric cut from the
inside by residual sharp edges.

Medium and large pneumatics need a more courageous approach, otherwise
known as brute force.  Safety precautions are imperative.  The
technique I use involves the use of a propane torch, with the smallest
flame possible.  It does not take too much trial error to get a feeling
for how much heat to apply against the old fabric.  It will get
scorched, and will be destroyed, so good idea to take span measurements
and write them down before removing the old fabric.

While constantly moving the flame along the glued surfaces, I quickly
heat up the panel such the old glue actually breaks down and
crystalizes.  It will soften quite quickly, and the fabric can be
removed fairly easily.  At this point, there is still residual animal
hide glue left behind on the bare wood.  Sanding this is an exercise in
futility.  Furthermore, the friction of sanding warms the old glue
which quickly makes the sandpaper useless.

I hit the residual glue with the moving naked propane flame, watching
carefully for crystalization, which is usually real quick.  Use extreme
caution here for it very easy to overdo it and scorch the wood.  The
crystalized glue then can be removed super easy with medium sandpaper.
It literally falls off in a dust, leaving behind clean wood, ready for
fabric replacement.

I would suggest practicing this technique on old scrapped out player
piano pneumatics that are headed for the junk box.  A certain skill
level is needed to get it right, and the eventual process is quite
quick and reliable.

I do have a heat gun that I have used from time to time, but this slows
down the process enough that I keep going back to the propane torch.

Safety precautions are very important.  Do have nearby a shop fire
extinguisher, full and certified.  Do have nearby a pail of water and a
wet rag.  From time to time in the process, the old fabric will
spontaneously combust into raw flames, upon which the wet rag and pail
of water are handy.

And finally, place the old fabric, which is certain to be scorched in
many places, into a metal pail, and leave it outdoors overnight.  Do
not put this fabric into your shop garbage box.  There may be hidden
glowing embers in some of the fragments.

If the weather is decent, do the entire process outdoors.  If you do it
indoors, risk of fire is always present; presume the worst and take
reasonable precautions.  And the smell throughout your entire home is
dreadful, certain to provoke domestic disharmony.

I share this technique with caution.  I'm aware there are a number of
people around who use similar techniques, and so far I'm unaware of
serious problems emerging from the process.  There may well be others
out there who have developed other approaches that achieve the same
result.  While this process works for me, and is reasonably swift, I'm
always receptive to another approach that may be quite different, but
equally effective.

What are others doing?

Regards,

Terry