[ Ref. Peter Phillips in 191013 MMDigest ]

Regarding Academia and Piano Rolls, I choose to distinguish between
public and private spheres.  The public sphere, be it academe, museums
or whatever, has the essential nature of being formalised with funding,
stated goals, policies and a governance mechanism to deliver the goals. 

The public sphere delivers continuity, procedure and a disinterested
community who apply all their academic thinking to all subjects.
('Disinterested' means they have no interest in any particular finding
but follow the data and their professional practices.)  Which is why
universities have followed their own path regarding piano rolls.

Private individuals most obviously can't provide continuity; sadly, we
read all too often here of unique material being lost.  They do things
for their own personal reasons and in any way they care to.  They bring
great knowledge, skill and hard work to these ends.  They are frequently
far from 'disinterested' but they have great interest in proving
particular outcomes.

Hobby societies sit somewhere between the two -- they are essentially
individuals clubbing together, but with some continuity and formality.
And also social media (such as MMDigest and Facebook).

I see all of these spheres as complementary, each feeding off the
strengths of the others.

It's obvious that in, say, 50 years there will be far few working player
pianos than now.  But, the intellectual property embedded in rolls can
be re-used by other devices indefinitely; these days the devices are
solenoid pianos and computer simulators, but again in 50 years who
knows.

I beg to differ with Peter Philips on how to achieve this; optical scans
are the only way to go.  A full-image optical roll archive records all
the data for most core aspects of the original physical roll.  This is
basic archiving: you record everything about the artifact.  A partial-
information archive has by its very nature limited use.  That's the most
obvious reason why the academic world has chosen advanced optical
imaging, and will continue to do.

Images can be used for all the classic physical analysis of rolls,
such as the punch patterns that are signatures of the roll editing and
manufacture, plus any printing or marks on the roll.  And images allow
that rather special property of rolls -- that they encode musical
performances -- to be used.  The image can be processed into a replica
master, and that can be used just like the original roll.  If you have
a replica master you have perfect and full knowledge of the original
performance.

Peter Phillips says "MIDI files derived directly from roll scans do not
preserve a number of factors such as paper acceleration, tempo and note
length."  This is an interesting assertion, but surely wrong.  It is at
best a generalisation about a complex and nuanced issue.

Hand-played rolls (mostly) embed in their perforations compensation for
acceleration during playback -- achieved by accelerating the original
recording roll.  Replicated masters contain precisely what's cut into
the roll, so simulation software must clearly behave like a player piano
by accelerating the speed with which it plays the master.  And every
emulator I know of does this, carefully and in a configurable manner.
And of course it's easy to measure or calculate tempo and note length,
either from the master or the (accelerated) simulated performance.

I said this last year at the Global Piano Roll Meeting in Leipzig,
and demonstrated my Welte simulator to give an idea what can be done.
The forthcoming event in Bern allows much more time and it'll be
interesting to see what topics get onto the agenda.

Julian Dyer

 [ http://www.hkb-interpretation.ch/2nd-global-piano-roll-meeting.html 
 [ "The 2nd Global Piano Roll Meeting will take place as an
 [ international conference in Switzerland at the Hochschule der Künste
 [ Bern and the Museum für Musikautomaten Seewen June 6-8, 2020.  The
 [ conference will include papers, presentations, and performances."
 [ -- Robbie