Re: Humming, DSP & musical reductionism

[John Hill <jphill@home.com>]

Baldwin, Daniel wrote:
> 
> I have begun to acquire some CDs on labels such as Pearl (Edwin Fischer's
> recording of WTC), Dutton Labs (Beecham's recordings of Mozart symphonies)
> and Arkadia (an Italian label-- I just purchased Cortot's recording of the
> Chopin etudes) -- all of these are transfers from 78s. The Dutton and
> Arkadia recordings are remarkably free of surface distortion; they have a
> kind of unearthly purity which almost sounds real, but not quite.

This can happen when *too much* of the surface (and other) noises are
removed.  Since the ambience of the original recording is buried down there
in the noise floor of the original 78, it too will be removed when the digital
algorithm for the de-noising is employed.  "Unearthly purity" is a good
description, because real pianos in real rooms or halls just don't sound
like that.  The brain knows this inherently and tells you that something is
not quite right when you have a recording that sounds like it was done
in a dead broom closet.

> There also
> seems to be excessive treble on the Arkadia recordings. On the other hand,
> the Pearls have more surface noise, but sound less artificially treated; it
> seems like I am getting a performance (under all the noise) that is truer to
> the original, but then again the sound is sometimes so muddy that it is
> difficult to make out the original.

This is where the decision-making process of the project engineer and editor
come into play.  Less treatment may give a better result, but there's no hard
and fast rule there.  On the Arkadia recordings, they may have tried to boost
lots of high-mid and high frequencies in an attempt to get more articulation
and less muddiness.  But it's a trade-off with no free lunch.  You might end
up with a brittle, harsh-sounding piano tone, which is strident and fatiguing
to listen to.

This points out the fact that older recordings (like the ones you're mentioning)
have *lots* of technical limitations associated with the first generation master.
The most common problems include limited bandwidth, limited dynamic range,
high noise floor and high distortion levels associated with the recording chain.
So there will be a limit in terms of how good a result can be obtained through
the restoration and remastering process.  It's not voodoo, but there are certainly
aesthetic judgements (important ones) that will have to be made by those doing
this work.

>Are my perceptions true to what you know
> about the processes used by these various labels? What transfers do you
> recommend? 

I think your judgements are right on the mark (although I don't have these
particular CDs to compare notes).
The question is whether you'd really prefer a straight-across transfer from
the original master format to a digital master, without any remastering process.
I doubt it.  Recordings of the vintage you're discussing tend to need lots of
assistance while making their journey to CD, IMHO.

>Does the process of "cleaning" and "restoration" yield a
> product which may be more pleasing to the ear, but is not worth the
> sacrifice, in that the end result is virtually a different performance? Or,
> are there sonic modifications typically introduced by the restoration
> process for which, once I know about them, I can make adjustments  when I
> listen, so as to filter them out?

The artifacts of restoration and digital remastering can be subtle, or they can
be really obvious.  In general, the sort of artifacts we're talking about (throwing
the ambience out with the bath water, etc.) are *not* things that you will able
to do much about at home with simple playback controls.

I think the idea, philosophically, is to get to the essence of the music.
Having to listen through layers of noise and distortion prevents one from really
hearing what the performer was trying to do.  This is particularly true, I think,
for critical listeners.  Others won't really care, 'cause they're probably listening
on a boombox or some other really compromised playback chain anyway.

> Incidentally -- what is your opinion of Sony's transfers of GG's recordings
> from analog to digital format? I recently read a review of the GG set in the
> Philips Great Pianists  of the 20th Century series (it is in the July/Aug
> issue of American Record Guide) opining that the engineering on the digital
> transfers of the Byrd and Gibbons tracks included in the set is far superior
> to the engineering of the same tracks on the Sony issue ("A Consort of Byrd
> and Gibbons"). What do you think?

I find them to be quite variable.  I generally like the Sony GG editions much
better than the Columbia masterworks versions that came out in the 80s.
And I've always liked the recordings made by Kazdin in the late 70s at the
Eaton Auditorium.  I used to have many of the vinyl versions of GG, but alas,
those milk crates are still in LA and I'm not.  It would be nice to do an A/B
comparison between the GG vinyl and the new Sonys.  I tend to think I'd
give the nod to the Sony GG Editions, but...

cheers,
jh