Dan Wilson's note a few days back (97.01.15) gave a good picture of the
Duo-Art seminar held at Denis Hall's last week.  I would like to expand
a bit on a couple of areas which MMDers may be interested in: how Duo-
Art rolls were processed.  They fascinated me, anyway!

First, Denis looked at the earliest Duo-Arts, which were retro-coded
from Metro-Art rolls which had no dynamic information from the original
recording (the catalogue shows this to be about the first 50 titles, by
the way).  I have one of these Metro-Arts and took it along.  Although
the performance note attack is the same, the Duo-Art has been signifi-
cantly processed elsewhere.  The extended note perforations in the
Metro-Art have been removed, and the soft pedal and themeing have been
completely redone from scratch.  We will be looking closer in this area
soon.

I pedalled my Metro-Art first, and it came out as a quite sleek
performance.  Nothing wrong with the notes.  When we put on the Duo-Art
-- Oh, boy!  It was dreadful and choppy, and all introduced during the
coding (if you hadn't the Metro-Art to compare you would thing the note
record itself was bad, but it is actually fine).

Intriguingly, Denis possess two copies of another early Duo-Art:
one is obviously from the Metro-Art, with extended perforations and
odd coding.  The other is the same pianist and number, and is a
completely different performance.  Aeolian simply re-recorded it from
scratch!

Denis then presented some information about how the later rolls were
converted from the note and crude dynamic information off the recording
console and reiterating perforator.  These were based on a sequence of
edited master rolls of a performance, all now stored in the Maryland
archives.  This is the process followed through most of the system's
history (for classical rolls; dance rolls were obviously different).

The 'original' roll (from the perforator) has been marked with pencil
dashes against the notes to be themed.  The dynamic codes, which are
all over the place because of the way the knobs were twiddled during
the recording, have been grouped into segments perhaps an inch or two
long, and a single dynamic level written in for that group.  An
artistic judgement has been made to do this.

The next version of the roll possesses completely new dynamics, exactly
following the pencilled-in values (not carrying any of the original
coding over).  Subsequent versions of the roll then add back details
until the performance is 'signed off' for issue.  The rolls are all
dated and labelled so it is easy to see how it has been done.

This shows an interesting process of recording in real time, editing
right back to give a simple (and therefore manageable) base set of
dynamics, and then selectively adding back detail until sufficient
complexity of performance is obtained.  Denis showed that it took some
years for this system to evolve, as earlier 'all the details' coding
took a lot of time and didn't work very well, presumably because the
coding information was simply too complex and confused to process by
hand.

What I found so convincing about all of this was that it referred back
to actual master rolls, which give physical evidence of the editing
process they have undergone.  Hopefully Denis will write all this up so
it can be presented to a wider audience; I feel all the evidence is
coming together and a convincing picture of a pragmatic but honest
recording process is emerging: one which combined artistic judgement
and technology in their best proportion.

The seminar, with its sequence of early-to-late rolls, and liberal
gramophone comparisons, was immensely convincing and a great de-
mystifier.  The Duo-Art really does come very close to the real thing
if given the chance!  Perhaps we could do it all again with a video
camera there...

Julian Dyer