I recently purchased a wonderful Chickering Ampico B and have been
enjoying it immensely.  It is a lifelong dream come true!

In addition to the satisfaction of playing the instrument and listening
to rolls, I have some rather crazy and naive ideas about incorporating
computer control into the instrument.  I have three goals for this
project:

    - Allow computer-controlled playback via MIDI files or similar;
    - Provide accurate electronic recording;
    - Create new rolls from digital/MIDI sources.

The recording device should be capable of recording live performances
to create new hand-played rolls.  It should also handle recording of
playing from existing rolls.  Combined with the ability to create
rolls from digital data, this should allow decent reproduction of rare
and/or damaged rolls without the fuss of optical scanning.

As a newbie to reproducing pianos and player pianos in general, I am
looking for some advice.  I can handle the electronics and software
design myself.  I need some additional expertise for the mechanical
pieces.  Ideally I will not be modifying the player mechanism itself at
all.  That is far beyond my capabilities at the moment.

                 PLAYBACK

For computer-controlled playback, I am aware of the Virtual Roll, but
the system is out of my price range and too difficult to install myself
anyway.  I propose to design a tracker bar interface, similar to past
attempts such as the PowerRoll.  I realize that the result may not be
as good as interfacing to the player mechanism directly but it is a
good start for me to learn about the technology, if nothing else.

Does anyone have recommendations on how to create the proper seals
around the tracker bar itself?  How about electrically- or
electronically-controlled valves?  Perhaps something like this:

http://www.anver.com/document/vacuum%20components/vacuum%20valves/elec_air_valvs.htm 

I have no idea what kind of flow rate a valve would need to activate
the piano's pneumatics.  Any advice?

                 RECORDING

For recording purposes I propose to use a force-sensitive resistor to
record the pressure on each key.  Ideally I'd build an array of these
to slide between the keybed felts and the keys themselves.  This way I
can capture expression as well as the notes themselves.  I was looking
at something like this:

    https://www.sparkfun.com/products/9673 

Unfortunately, this caveat applies: "These sensors are simple to set up
and great for sensing pressure, but they aren't incredibly accurate.
Use them to sense if it's being squeezed, but you may not want to use
it as a scale."  The manufacturer quotes a 6% force measurement
variation from device to device and a 2% repeatability variation on the
same device.  I may have to write some software to calibrate each
sensor individually.  That shouldn't be too difficult.

Has anyone worked with these or similar devices?  Any experiences or
suggestions of other sensors would be helpful to hear about.  I will
need something that can react quickly to record rapid repeats and such
things.  I don't know if the above sensor is capable of that.

                 ROLL CREATION

I would like to build a computer-controlled roll punch but have no
robotics or similar experience.  That stepper motors exist is about
the extent of my knowledge.  I have found several such devices on the
web but of course there are no specs or plans.  I am thinking of
creating an array of punches to quickly create rolls without the need
to move an arm laterally.  Hopefully only a motor to advance a paper
spool would be needed.

Are there any recommendations on what kind of mechanical punch to use?
Any (stepper) motor recommendations?  If anyone has plans for such a
thing that would truly be amazing.

Each of these projects seems pretty big and I don't expect to get even
one of them working without months (probably a year or more) of
experimenting.  I get a lot of joy out of tinkering with things so the
journey itself will be its own reward.

Thank you for any advice and encouragement, or even discouragement if
may plans are a bit too on the "mad" side of mad scientist.

David Greene