Re: arsclist Arhoolie Foundation - Digitizing Project

["Doug Pomeroy" <pomeroyaudio@xxxxxxx>]

A couple comments, with much info which everybody already knows:

Everyone is in favor of recording at the highest resolution, but it makes
sense to look at the matter carefully when dealing with old recordings. What
is the resolution of the source? Old 78 rpm shellac pressings can have very
good fidelity in the best cases (eg., some of Victor's Orthophonics), but
even in those cases the dynamic range is far less than the 95 dB provided by
a well made16 bit digital recording: you reach pure surface noise well
before you actually risk losing any recorded information. The queston of
how many bits, is a question of dynamic range only.

Similarly, the frequency range of old recordings is more than accomodated by
the conventional 44.1 kHz sampling rate. (Your indication of "96 mhz" should
read 96 kHz.) 78 rpm recordings made on the Western Electric cutters could
record up to 8 kHz by the late 30s and the range was extended to 10 kHz, and
slightly beyond, by the end of the 78 rpm era. Those who believe the highest
frequencies in such recordings cannot be fully and accurately captured by a
high-quality converter running at 44.1 kHz are mistaken.

Recording at an unnecessarily low level, does decrease the resolution of any
digital recording. Realistically speaking, if you are making transfers of a
large quantity of 78 rpm discs, there might not be adequate time to set the
ideal level - to activate all 16 bits - for each and every one and so, in
such a case, I can appreciate where having a higher number of bits would be
seen as a good thing in helping to preserve low-level detail in the source
recording. This argument is addressed by George Brock-Nannestad, in his
post:
 > Sadly, sadly, 24 bit is not 24 useful bits at all. You cannot record 8
> bit
> below maximum level (ca. 46 dB below) which would get you 16 bit - these 16
> bits amplified would not get you 16 bit which are as good as 16 bit A/D
> conversion from a proper 16 bit A/D converter!
However, he also feels impulsive noise (eg., "pops & clicks") should be
preserved without digital clipping, and we disagree about that. Such a
requirement definitely requires recording at what I would call unnecessarily
low levels.

IF digital audio is going to be subjected to multiple passes through digital
signal processors, and this includes simple gain changes, then this could be
an argument in favor of recording initially at higher-resolution. With
archival audio, however, this argument doesn't seem particularly relevant.

Tomlinson Holman, who is something of an expert, has commented that 96 kHz
sampling is unnecessary in *any* situation, and results in a tremendous
waste of storage space. (He feels the highest meaningful rate would be ? 60
kHz, and then only for the very highest-quality source recordings.)

Someone suggested using CEDAR processing at the time the preservation
transfers are made. This is definitely not accepted practice. Don't do it.
Make "stereo" (two-track mono) flat transfers.

If you do make flat transfers, you need a preamp designed to give such an
output (in "stereo"), and there are really none on the market today, except
for the wildly overpriced FM Acoustics (and the one mentioned by George
Brock-Nannestad), which also has other useful controls re: turnover and
rolloff curves. But it is not rocket science to modify a preamp to defeat
the RIAA curve, and this is what I would suggest. (Stanton used to make the
Model 300B Phono Preamp, which has an optional "flat" eq output.)

Doug Pomeroy   pomeroyaudio@xxxxxxx
Audio Restoration [CEDAR] & Remastering
-------
> Dear Arsclist,
>
> I just joined up on this list so I don't know what has been covered
> before. But I was hopipng that some of you could us out.
>
> The Arhoolie Foundation is about to digitize the Chris Strachwitz
> Frontera Collection (A large collection of vernacular Mexican and
> Mexican American Music) of 78s, 45s, LPs, tapes, cylinders and other
> source materials. We are working with the UCLA library and the Chicano
> Studies Department there.
>
> The first part of the project, the 78s, comprises about 80,000 minutes
> of music (about 26,000 78 RPM sides). UCLA wants high resolution,
> probably, 96mhz/24 bit. They would like us to put it on a removable hard
> drive and send it to them as we do it and they will store it on their
> server at UCLA. We will also keep a copy for the Arhoolie Foundation.
>
> We are looking for answers or opinions on the following questions.
>
> A. Capturing the analog sound.
> 1. What is a good currently manufactured turntable with variable speed
> that will play 78s?
> 2. What is a good currently manufactured pre-amp, preferably one with
> various EQ curves?
> 3. What is a good cartridge that accepts a currently made 78 stylus?
>
> B. The digitizing. We would like to use a PC instead of Mac.
> 1. What is a good model A/D converter?
> 2. What is a good sound card for the computer?
> 3. What program would be good to use to record and save the files?
> 4. What file format would be best to save these in? Is there a format
> that allows meta data to be attached?
> 5. Ideally, the Arhoolie Foundation would like to keep it's copy of a
> RAID server. This would require about 1.3 terabytes. Using SCSI drives
> this would be quite expensive. However, there are new IDE ATA 100 RAID
> systems that are coming about that are much cheaper and are large
> capacity. Do you have any knowledge about this.
> 6. We will also be taking a high resolution digital photo, of the source
> material. In the case of the 78s, this will be the 78 label. Do you know
> of a database system the would work as a good front end to manage all of
> this?
>
> Thanks forany help you can give. Please email or call if you have any
> questions.
>
> Best Regards,
>
> Tom Diamant
>
> Arhoolie Foundation
> 10341 San Pablo Ave.
> El Cerrito, CA 94530
> USA
>
> phone: 510-525-7471
> fax: 510-525-1204
> email: tom@xxxxxxxxxxxx
> web site: http://www.arhoolie.com