Regarding "pay the mortgage on the laser machine!"  Here's where
I have to beg to disagree.  I've been working on this area too
because there has been a big change: the cost of a laser cutter has
fallen a great deal.

The bigger ones owned by hackerspaces and maker centres cost a few
thousand, but we don't need a large bed size for standard rolls.
Low-end Chinese clones with a 40-watt gas laser (which will cut about
6 mm plywood, will make short work of music rolls) and a 12" by 8"
bed now cost new £330 on eBay ($500 in the US?).  Not 'cheap' cheap,
but not really mortgage territory; if you can afford to have a player
piano delivered and buy the materials to refurb it, you can probably
save up for your own laser cutter now.  This is a huge change from
five or ten years ago.

Admittedly, they're a bit rough and ready at that price, and it takes
some knowledge to get them to work well.  For roll cutting and driving
a take-up spool we would want to change the Chinese control circuits
for open source RAMPS circuits -- call it another £30.  A diode laser
might even do the job, and be even cheaper to get set up with, but the
ones I've seen seemed marginal, whereas the K40s and their ilk can be
quite good.  I'm snarled up in house renovation at the moment, but as
soon as I get somewhere to put one I'll try to provide a hands on
review.

Although laser cutting can be pretty fast, and for rolls we will need
the lowest powers and highest speeds the machine offers (you need to
spend greater than £1k to get serious max speed and a lot of small
rounded holes is a slow pattern), this plan probably won't ever cut
serious commercial quantities but it should be enough to fill a useful
niche.

I see the way forward with affordable lasers as being to get "fast
enough" cutting into lots of hands cost-effectively, rather than chase
mass production for a few bulk producers.  I'm hoping it will be
feasible to produce a conversion kit: buy the kit, buy an eBay laser,
follow the instructions, cut rolls!

I've written a prototype program that reads MIDI and generates an
SVG vector graphic of a piano roll, which is a good starting point
for laser cutting.  If I test it and it works, I'll open source it.
(The SVGs look right-ish.  I've not cut one yet; maker spaces look
very nervous when you say you want to modify their cutters to hold
large rolls of paper).

By the way, the scalpel-based cutting plotters (vinyl cutters) are
not a good bet -- I already looked at that option and gave up.  The
cost is a substantial fraction of the laser anyway, though admittedly
the care and feeding requirements are fewer.

Lester Hawksby, UK