[ARSCLIST] A proposed A/B test (was "dumbing down" of Downloaded Recordings)

[Jon Noring <jon@xxxxxxxxxxx>]

Clark wrote:

> Just to add some accelerant -- I've never (yet) heard an LP or a CD made
> from 78s that portrays the range and beauty of the originals as I play them
> on my own system. (One of my special tricks: application of high-end audio
> technology and practices.)

I've heard this from several 78 collector friends I've known,
including one who I regard as a true audiophile.

I think a lot of this "effect" is due to the cleanup/restoration of
the raw transfer, and not fully understanding the psychoacoustics of
noise in our perception of the signal during cleanup. (Noise,
particularly truly random "hiss", has the effect of "brightening" the
audio as humans perceive it. This effect is also seen visually with
graphics images, interestingly enough -- at one time I had a link
demonstrating this beautifully, but can't find it. :^(  )

I will assert that we cannot tell any difference between playing
back a 78 live, and playing back a digital rendition of it from the
same playback system, if:

1) the AD and DA steps are professional grade, 96k/24-bit or better
   sampling, and

2) we do not post process the raw digital transfer -- we keep it raw
   and untouched per the original.


This is why I highly value very high quality transfers and
digitization, and that such raw transfers are archived with no
post-processing of any kind. That is, the focus for archival purposes
should be on getting the absolute best transfer, and very high grade
digitization, and do not do any audio cleanup/restoration -- store it
"raw."

This does not mean the raw transfers should never be "cleaned up" by
those interested in doing so. But we should never throw away the raw
transfers (so long as they are properly done).

It would be interesting to maybe setup a blind A/B comparison to
verify my assertion. Get several "audiophile" and 78 collectors
together to see if they can tell the difference between:

A: 78 playback --> pre-amp --> amplifier --> speakers

B: 78 playback --> pre-amp --> AD --> DA --> amplifier --> speakers

Provided of course the AD/DA is done at 96k/24-bit sampling using
the highest professional grade equipment, and all other equipment
remain the same. If so, I assert the blind test will show no
discernable difference, and if this is indeed the case, then simply
putting a digital storage step between the AD and DA steps will not
change the "calculus" since the storage does not alter the bits one
iota provided storage is lossless.

Now maybe I have put too much faith in high grade AD/DA equipment.
Let's run the test and find out! Anyone with golden ears willing to
take part in such a comparison test?

Jon Noring


(p.s., since I know someone will ask why 44.1k/16-bit is not sufficient.
What we really need to do is to accurately represent the noise in the
recording, especially impulse type noise, and this does require a
higher sampling rate and bit depth to get a fairly good representation
of the noise. The analog side of the AD clearly has to be pretty
accurate on impulse type noise. I'll let Erik and others common on
this since I know little about the real world of state-of-the-art AD
converters.)