The reason that any spring-operated regulator plays too loudly is that
the spring tension is too high for the given expression coding.  In a
Duo-Art, it's easy to get the zero level right, but if the spring has
too high an initial tension the expression will pull away too fast as
the spring is stretched.  Aeolian said that the spring should be "just
loose" at rest, or something like that.  That means the initial tension
in the spring is nil.

When the regulator closes and stretches the spring, the idea is that
the stretch from rest to zero position will give the correct tension
for the zero level.  The approach I've adopted is to alter how far the
regulator rises up to zero level.  Although this is stated as being half
an inch, there is range in the system to make this a bit more or less.
If you want the dynamic range to be a bit lower, fitting a weaker
spring and having the regulator rise a bit further to zero will do the
job nicely.  This is more in keeping with the design of the system than
reducing accordion travel.

Spring tension goes up as the cube of the wire gauge, so switching
between available standard springs of 0.047 and 0.055" is a large leap.
A more subtle way to reduce spring rate is to increase the free length
of the spring, such as by extending the bracket that anchors it.  A
friend did this successfully with a Steinway grand in a small room.
The additional piece just bolted to the end of the T-piece that holds
the springs, and could easily be removed.

Julian Dyer