Re: [ARSCLIST] Certification (was Re: [ARSCLIST] Wire recorders)

[Jon Noring <jon@xxxxxxxxxxx>]

Richard L. Hess wrote:

> There seems to be an interesting dividing line between the straight
> transfer and the cleaned (edited) copy.
>
> I insist that the archives that I work with take two unprocessed
> transfers and then I also provide (typically) two processed copies.
> Only the most enlightened archives with good material seem to
> understand this.

The proposed approach for Project Gramophone (Sound Preserve), is to
create high-quality *raw* digital transfers of the original 78 source
material, and to make them freely available to the public in some
fashion, as allowed by law. (How to meet legal requirements is being
investigated by Brewster Kahle at the Internet Archive who has already
consulted with IP attornies in Europe -- waiting to hear back about
his findings.)

The rationale is that over time digital audio restoration technologies
will continue to improve and evolve, so having the raw, high-quality
digital transfers archived and available is more important. That is,
once a raw transfer is digitally restored, it can't be returned to the
original raw digital state for new restoration -- a restored version
is, in effect, a third-generation copy of the original (1. original
disc --> 2. raw digital transfer --> 3. digitally restored copy.) It's
much better to have a digital library of raw transfers rather than the
restored versions where the raw transfers are not preserved or made
available.

Of course, one can argue that the technologies for digital transfering
from 78 source material will also continue to improve. And this is
true. However, there is a need for digital preservation NOW, as well
as into the future. So the key is to do the absolute best transfers,
within reason of course, today, using today's technology (which is
both stylus playback and optical laser playback -- I'd try to do both,
when possible, for the same disc.)

As far as I can tell, the ultimate long-term raw digital transfers
will be done using ultra-high-resolution optical microtopography (which
is now being studied at Lawrence Berkeley Laboratory.) But unless there
are some unforeseen breakthroughs, I don't see that as being practical
for at least decade (and probably more) except for extraordinarily
rare and noteworthy discs and cylinders (the scan time has to greatly
speed up, and the amount of digital data collected per disk will also
be huge, probably many times more than high-quality digital transfers
using today's playback technologies.) Some of the intermediate-level
optical topograhic scanning techniques LBL is now working on might
become competitive with mechanical/laser playback a little sooner --
when is hard to say. But again, we can't predict if the LBL research
will ever lead to a practical digital transfer system, so no use
waiting around for a couple decades to see if that will lead to a
much-wanted breakthrough.

In the meanwhile, for digital archiving *today*, mechanical and laser
playback is the state-of-the-art. As I see it now (subject to change
as the experts converge on the requirements), for "regular" 78 discs,
use a single playback speed of 78.23 (unless the recording speed is
known with 100.00% certainty), use a totally neutral (flat)
equalization (system calibrated). Digitize with professional-grade
equipment at a minimum of 96k/24-bit/2-channel resolution. Use a
top-grade turntable/tone-arm/cartridge combination (as well as that
Japanese laser turntable whose name eludes me at the moment.) Do maybe
2-3 passes with different styli diameter/shapes (and if laser playback
is possible, do that, too.) Of course, use the best system to clean
the discs when allowable. Record the full metadata of the performance
as given on the record, and digitally photograph/scan the label,
run-out area and anywhere else where data occurs on the disc. Record
the full metadata regarding the transfer process itself. I could go
on, but this covers the major items I can think of from the top of my
head.

Jon Noring
Project Gramophone