There have been various references to using the scalloping down the
edges of slot perforations as a means of re-discovering the actions of
the original perforator.  I think it's most unlikely that anyone will
ever develop a system based on this approach, nor is it necessary.

The essential requirement of a roll reader good enough to be able to
re-create the master is that it can accurately measure the start and
finish of every perforation on the roll.  If the perforation was a
single punch, the difference between start and finish gives the punch
diameter.  It the perforation was a slot, it would have been created
by a series of consecutive punch strikes.

How many?  To find out, measure the total length of the slot, subtract
the punch diameter, divide the remainder by the perforator step dis-
tance and there's the answer. The joy of this technique is that it's
digital.  It's either right or wrong, there is no middle ground.

The underlying software techniques are relatively straightforward,
although special provision needs to be made to control the tracking of
the reading process, both horizontally and vertically, because of warped
paper.  My approach employs a predictive technique based on calculating
a notional anticipated position of each event read from the roll and,
by comparing that with the actual result, generating error data which
is fed back into the loop.

I assume this is standard practice and gives rise to the term "phased
locked loop", inappropriate in this situation because most engineers
apply that term specifically to oscillator locking.  There is no
oscillator involved here.

Richard Stibbons

 [ The spatial frequency of the perforations is quite steady in a piano
 [ roll, and so the phase-locked-loop (PLL) technique works very well
 [ to recover the "local oscillator", which is the constant spatial
 [ frequency created by the paper advance mechanism in the original
 [ perforator.  The PLL circuit which maintains the phase of the color
 [ sub-carrier oscillator in a TV set is related.  The technique is
 [ the same, irrespective of its application, whether it is applied to
 [ a periodic event in time (an oscillator), or to a recurring event
 [ along the length of a piano roll (the periodic advance).  Whatever
 [ the name, the technique works very well when transcribing piano
 [ rolls.  -- Robbie