Re: [ARSCLIST] current role of soundcards

[Jon Noring <jon@xxxxxxxxxxx>]

Sam posted:

> An informal, unofficial notice:
>
> Earlier this year, the National Recording Preservation Board
> commissioned a set of white papers relating to digital-related topics
> in sound preservation. The very first of these papers to be
> "published" was placed on the web this very morning. It is
> "Measurement and Evaluation of Analog-to-Digital Converters Used in
> the Long-Term Preservation of Audio Recordings" by Ken Pohlmann.
>
> A formal announcement of this publication will come from LC in the
> next few days but given today's thread, why wait? It's a wonderful
> work.
>
> http://www.clir.org/activities/details/AD-Converters-Pohlmann.pdf

Great!

Although I've not yet carefully read the paper, a skimming through of
it concludes that even for old material, high quality equipment and
digitizing at 96k-24bit or even 192k-24bit should be done.

One curious thing that I did not see noted in the paper is the need to
not only transparently transfer/digitize the wanted audio signal, it
is also important to transparently transfer/digitize the *noise*.
(Eric Jacobs first brought this up, I believe.)

Getting an accurate picture of the noise is just as important since it
may aid in the restoration process. Since the noise played back from
old recordings can go into quite high frequencies (e.g., due to high
Fourier harmonics found in impulse noise), to accurately represent it
alone suggests 96k or even 192k sampling. For grooved recordings, it
also suggests the need to get the lowest inertia cartridge possible,
which I assume would be moving coil (or laser). This should, I
surmise, minimize the "width" of any impulse noise in the groove,
closer to the theoretical minimum. Make the impulse noise narrower,
the easier it is to recognize and remove.

(As I think about impulse noise caused by "boulders" or "holes" in the
groove, the bigger the stylus, the wider timewise will be the noise
impulse. This is due to simple geometric considerations -- it suggests
for groove recordings that the width of the stylus near to where it
contacts the groove *in the direction of travel* should be minimized
as much as possible. Combined with a moving coil setup, and 96/24 or
192/24 digitizing, theoretically should result in the highest quality
representation of the noise, and greatly aid in restoration. Of
course, that's my theory... <smile/>)

Jon Noring