Pitfalls of Comparing Old Records to New Digital Transfers
8-H Haggis
I have decided to suggest a new thread, based on
some of the comments made in the discussion of the
Reader's Digest vs. Chesky vs. Chandos issues of
the Wild/Horenstein Rachmaninoff Piano Concerti
sets, and broadening the topic to cover the
comparison in general of old records with their
recent CD transfers.
This enterprise often involves the classic
situation of comparing apples to oranges. Here's
why:
1. Phono equipment. Record company or radio
station engineers (such as my former self) will
know that it is nearly impossible to obtain truly
"flat" disk playback equipment, which has uniform
frequency and linear phase response, and
consistent stereo separation across the audio
spectrum.
For many years, I worked with radio stations as an
"audio consultant" before the days of CDs. Many
program directors were concerned that another
station in the market might have a very different
"air sound": one which was more spacious, or had
more highs or lows, or had a juiced-up, glamorized
quality.
I would typically begin with the phono cartridge.
Employing a variety of test records (such as the
series made by CBS Laboratories), I would find
that the 'broadcast cartridge' (a heavy-duty model
that would withstand back-cuing) had deficient
high frequency response; this would require tuning
the input of the phono preamp with capacitors,
resistors, or in some extreme cases, repairing a
defectively-designed preamp that did not comply
with the RIAA curve.
On the other hand, it was usual to find that home
audiophile cartridges tended to have BOOSTED high
frequency response: for example, the original
model of the Shure V-15 series, had an
extraordinarily "sweet" high end, with very large
increases in output nearing the resonance point of
the mechanical/electrical assembly: this caused a
rise of many dB at, say, 15,000 Hz. If this
cartridge were used to play records on an FM
station, illegal overmodulation would result, or
the station's limiter would be severely
overdriven.
During the heydays of stereo LPs (and the "Shaded
Dog" era and shortly thereafter) these and similar
kinds of phono cartridges were the standard
pickups with which records were enjoyed and
compared.
Furthermore, phono cartridges are notorious for
having inconsistent stereo separation across the
audio spectrum. At some points in the midrange,
the separation may APPROACH 30 dB; but at certain
low frequency regions -- nearing a mechanical
resonance point of the cartridge/stylus
assembly/tone arm/record platter system --
separation may fluctuate wildly. And at high
frequencies, it may dip to just a few dB. Of
course, in contrast, modern PCM recorders have
very wide, consistent frequency/phase response and
stereo separation across the audible spectrum.
To really hear what is in the grooves of a
phonograph disk, it is necessary to use a KNOWN
cartridge and matching preamp. One must not
merely 'like' the sound; one must verify with
instrumentation -- averaging the frequency
response curves using several test records -- of
the playback equipment. The golden-eared
audiophiles of the "no measurement" school would
become apoplectic at this suggestion: they believe
that such tests are irrelevant.
If they ARE: then I have a question. Why is it
that the original engineers for the recordings
they admire and love, always used these
conventional techniques to calibrate and tune
their recording chains?
2. Original Mixdown Practices. I discussed at
great length in a previous post the mechanical
artifacts of the Westrex stereo recording/playback
process: the phase nonlinearities, the mechanical
and electrical filtering, the low-frequency
separation compromises, etc.
Now, one should also consider the aesthetic
considerations of the original mixdown/cutting
engineers. Recordings of the "Shaded Dog"
variety, virtually all Mercury Living Presence
albums, Columbia 360-Sound Stereo disks, and
others that were highly regarded during the late
50s and early 60s, were done typically on 3-track
Ampex machines (300s, 351s, etc.) using 3 basic
mixing channels.
The 3-track tapes were conventionally mixed to 2
channels for a disk-cutting master. It was up to
the aural concept of the producer at that point to
determine the relative balances of the channels:
this could have a tremendous effect on the body of
the recording, the "width" of the sound-stage, and
the instrumental differentiation.
If the original mike placement of, say, a
primitive 2-track "Blumlein" recording (such as
the early EMI stereo disks of Klemperer and
Beecham) were not mixed just right, the hall
balances and instrumental separation could be
insufficient (too little L-R); thus, some early
stereo pressings of these recordings seem rather
anemic compared to later and more careful mixes as
preserved on recent CD transfers.
Since we live in the "real world", there is little
chance that an engineer in 1997 could produce, by
ear, the PRECISE balance chosen by another
engineer in 1959 for producing a disk-cutting
mixdown. One can only aim to get close, given
that the original LP disk is believed to present a
good reference.
In some cases, early stereo disks were NOT very
good: I owned many early FFSS disks by Ansermet
(such as Rimsky's "Christmas Eve" Suite and
Dubinushka) that had almost NO separation. Too
little L-R was mixed in; later CD transfers have
opened up the soundstage and present antiphonal
effects that are only mildly suggested in old LP
stereo disk pressings.
Passive audiophiles who have NEVER had the
opportunity to personally adjust a mixdown of
Blumlein-type recordings (L+R on one channel, L-R
on the other), or 3-channel masters, cannot begin
to understand the extreme sensitivity of the
process, and the subtle ways that the soundstage
and dimensionality of the recording may be
affected: a level shift of a couple of dB may make
all the difference in the world!
3. Peak Limiting and Its Subtle Enhancements of
Detail
We all have heard about "peak limiting" but only
an audio engineer who has used such devices can
demonstrate the subtle (or gross) effects they can
produce.
As Tony Movshon has commented on this newsgroup, a
bit of peak limiting, tastefully employed, can
confine an otherwise impractical dynamic range
into a convenient package for home listening.
The audio engineers who produced these grand old
LPs of yesteryear were like Michelangelos or
Titians, compared to the ham-fisted clods who
populate the engineering staffs of radio and TV
stations! They knew that 2 to 5 dB of judicious
limiting here or there might make all the
difference in the world; might mean that 100%
rather than 85% of phono cartridges could cope
with the groove modulations; might make the
recording sound "big" and "grand" rather than just
accurate.
One can take a superb mastertape -- such as Layton
typically produced for Reiner or Munch -- and then
peak limit one or two spots as needed. But that 6
dB slice off the top of a climax -- though not
excessively audible at the moment -- has an
interesting side-effect.
Now, all the softer passages are, ipso facto, 6 dB
louder. The "inner detail" is more clarified.
Hall resonance is just a bit more pronounced. All
throughout the entire recording!
Sometimes the peak limiting was employed during
the session. Almost always, it was also ladelled
on during the mixdown or disk cutting process.
If one wishes to perfectly accurately compare an
original orchestral master tape recording to a CD
reissue, one has to:
1. Have the original peak limiter in hand;
2. Duplicate from session notes the compression
utilized in each stage up to disk-cutting;
3. Duplicate any sweetening equalization used
at any step of the process.
4. Compare the equalized, peak-limited tape to
the CD that was prepared from it and to an
original LP copy.
But -- in our "real world" -- NOBODY does that!
We just plunk the record on our (uncalibrated)
turntable, listen, and then turn on the CD. What
we hear is assumed to be the final resolution of
the issue. But it is NOT a controlled, informed,
or accurate comparison at all.
4. Some Examples That Will Demonstrate
Phenomena Discussed Above.
a. Peak Limiting On Original Session Tapes.
Play back the RCA - BMG CD pressing of
the classic record THE REINER SOUND. Listen to
Rachmaninoff's "Isle of the Dead" or Ravel's
"Rapsodie Espagnole" and carefully listen
for the "squashing" of climaxes.
Note, for example, the incredible
definition of the inner voices and soft passages
in Ravel's "Prelude a la nuit", or "Habanera".
But then take careful note of the fact that in the
subsequent outburst in "Feria", the climaxes are
squashed and swallowed up. The bass drum punches
down the overall amplitude: a sure cue that a
simple broadband peak limiter has stomped on the
mix levels. Compare the original LP in either
mono or stereo, and you will hear this SAME peak
limiting and artificial-enhancement of soft
passages. Ergo, the limiting was probably in the
master-tape, since ALL subsequent copies replicate
it.
b. Poor Stereo Mix of Channel Separation or
Spread on LP Stereo Disk.
Listen to the original FFSS disks of
Rimsky-Korsakov's "Dubinushka" with Ansermet/OSR.
Note that the recording, though labelled stereo,
has very little separation. Compare the CD
transfer on London 443 464-2. The 1958 recording
is not the last word in perfection, but in the CD
playback it has more hall tone, a more solid bass,
and greater width.
c. Thin, Wiry Sound of Original LP is Tamed on CD
Audition the critically-acclaimed
Bernstein LP recording of the Haydn Symphonies 82
& 83, a classic performance that was among the
first of its day to employ certain performance
practices of the period that were being promoted
by scholars. Note that there is terrific
end-of-side distortion, "needle talk", and heavily
boosted, almost shrieking highs.
Compare the CD transfer in the CBS/Sony
Royal Edition of the Haydn "Paris" Symphonies.
The string partials are now sweeter, cleaner, even
more extended; there is greater breath around the
ensemble; deeper bass; richer lower-middle; and a
total lack of mechanical tracking distortion.
Why? Because the set is truer to the (excellent)
master tapes than the old LP pressing.
5. Conclusion.
All rules, it is said, are made to be broken. We
can find exceptions to any example (such as the
above) that may be offered. They do not disprove
the overall truth of the theory.
Before you assure me that I'm all wet, I would
hope that you speak from a position of strength.
Tell me that you are listening to the original
master tape of the recording, which has been
kindly lent by the company, in your
properly-conditioned music room.
Tell me that the original engineers or producers
have assisted in setting up the playback.
Tell me the calibration of all of your equipment,
and the experience you have had in making
recordings of classical music.
These are not nonsensical suggestions: such
conditions typically exist when the classic
recordings are remastered for CD.
I CAN be persuaded that I am utterly off the mark,
but it will require some effort to convinece me
that your "passive" comparison is as valid as the
yeoman-work of the technical staffs of RCA, Sony,
London, or Mercury in transferring their great old
masters to digital.
Yours,
8-H Haggis
8-H Haggis
[NOTE: I tried posting this some hours ago, but it
has not yet appeared on the server. If it does
turn up, I shall delete this repost.]
5. Conclusion.
but it will require some effort to convince me
8-H Haggis
eac...@aol.com
In article <3323183...@nntp.ix.netcom.com>, ***NoSPAM***pian...@juno.com wrote:
>Now, one should also consider the aesthetic
>considerations of the original mixdown/cutting
>engineers. Recordings of the "Shaded Dog"
>variety, virtually all Mercury Living Presence
>albums, Columbia 360-Sound Stereo disks, and
>others that were highly regarded during the late
>50s and early 60s, were done typically on 3-track
>Ampex machines (300s, 351s, etc.) using 3 basic
>mixing channels.
Thanks for your fascinating essay! I'm in no position to judge here,
but I must say that, whenever I play *any* of Columbia's "360
Sound" series LPs, I have to turn the treble all the way down
and *still* there's too much treble! The highs screech unmercifully!
(I have several Ormandy and Szell items in that category.)
Consequently, I hate "360 Sound" passionately! (Some background:
I have, since 1989, been hearing Dallas SO concerts in the
acoustically magnificent Morton H. Meyerson Symphony Center.
Prior to that, I heard that orchestra in a badly refurbished McFarlin
Auditorium on the SMU campus and in the sonically abysmal
Fair Park Music Hall (the latter is fan-shaped!). And, during three
years in graduate school at the Univ. of Pa., I heard the Philadelphia
Orchestra in the Academy of Music, the acoustics of which have
seemingly been "controversial" recently, but which I found dry
but not lacking in presence.)
FWIW, one of the very worst-sounding recordings I have ever
heard is Eugene Istomin's of the "Emperor" concerto (mono)
with Ormandy and the Phila. Orch. *Nothing* I try on the reproduction
will make it sound even listenable! A pity... --E.A.C.
Tony Movshon
pian...@juno.com (8-H Haggis) writes:
[snip] of a good chunk of very informative material
>3. Peak Limiting and Its Subtle Enhancements of Detail
>
>We all have heard about "peak limiting" but only an audio engineer who
>has used such devices can demonstrate the subtle (or gross) effects
>they can produce.
Among many other interesting points, your post raises a question about
the uses of limiting that interests me in the context of making
digital transfers of analog material. The question has several parts:
1. My impression is that peak limiting on the master tapes was rather
variable. Some companies (e.g. RCA) seemed to do more of it than
others. How much limiting on the masters was typical?
2. Conversely, it seems that everyone applied quite a bit of limiting
in the preparation of LP masters, in large part to make the disks
playable on the kind of equipment that a normal listener could be
expected to use. Some companies (Columbia, certainly) used to squash
LP dynamics mercilessly, a practice that has not generally been
continued in their CD remasterings. How much limiting was typical in
the LP mix?
3. (really 1+2) It stands to reason that limiting on the master tape
is difficult to overcome, but limiting on the LP is irrelevant
nowadays, unless you're unfortunate enough to be listening to a master
made from an LP. Can anyone (8-H?) give some reasonably precise
account of where the limiting in most analog material was implemented?
4. Finally, even if there is limiting on the master tape, it seems to
me that a diligent engineer armed with good software could make a stab
at "unlimiting" the tape, i.e. cranking the gain to undo at least the
frequency-independent components of limiting. I realize this would
probably be difficult to do on a regular basis, since the effect of
the limiter would be difficult to reverse-engineer, and the effect of
any manual "tapering" by the original engineer would be very difficult
to deal with. Still, for companies like DGG that seemed to rely on the
automatic limiters almost exclusively, it should be possible. My
listening to remastered DGG efforts suggests that they have not, at
least until now, tried to do this. Why not?
Tony Movshon
Center for Neural Science New York University
http://www.cns.nyu.edu mov...@nyu.edu
Tony Movshon
rsfo...@inforamp.net (Richard S. Foster) writes:
>On Mon, 10 Mar 1997 00:05:52 GMT, pian...@juno.com (8-H Haggis) wrote:
>
>>In some cases, early stereo disks were NOT very
>>good: I owned many early FFSS disks by Ansermet
>>(such as Rimsky's "Christmas Eve" Suite and
>>Dubinushka) that had almost NO separation. Too
>>little L-R was mixed in; later CD transfers have
>>opened up the soundstage and present antiphonal
>>effects that are only mildly suggested in old LP
>>stereo disk pressings.
>
>This is so wrong it goes passed misleading. I have enjoyed most of
>your music related threads, but this is a very foolish statement. I
>would invite you to come and listen to these Ansermets', the Decca
>versions, not the Londons, which, regardless of what *experts* have
>said, are do sound different.
Since almost all London records were pressed from the same masters in
the same plants as the Decca equivalents, this is pretty hard to
believe.
mike gray
Tony Movshon wrote:
>
> pian...@juno.com (8-H Haggis) writes:
>
> [snip] of a good chunk of very informative material
>
> >
> >We all have heard about "peak limiting" but only an audio engineer who
> >has used such devices can demonstrate the subtle (or gross) effects
> >they can produce.
>
> Center for Neural Science New York University
need to be corrected.
First, pace the ears of 8-H Haggis (whose radio experience isn't transferable to what he
hears on records), no recording engineer that I'm aware of ever used limiters on
stereophonic recording sessions. What you hear on many "Golden Age" master tapes is
tape saturation, a common pre-Dolby, pre-201 bending of a tape's transfer function
used to achieve the best signal-to-noise ratio.
Second, it's not true that the Blumlein EMI system had no way to alter stereo width on
the mixer - it did: otherwise the stereo engineer had no way (other than mike placement)
to achieve the right balance between left-right width and proximity to the orchestra.
Finally, I urge readers of this thread to take a look at my articles in back issues of
The Absolute Sound: all these subjects (and many more) were considered there with what I
like to believe was historical accuracy, something comments on this thread could a dose
of.
Regards,
Mike Gray
Classical Music Editor
Fi Magazine
Tony Movshon
mike gray <mhg...@worldnet.att.net> writes:
> First, pace the ears of 8-H Haggis (whose radio experience isn't
> transferable to what he hears on records), no recording engineer that
> I'm aware of ever used limiters on stereophonic recording sessions.
> What you hear on many "Golden Age" master tapes is tape saturation, a
> common pre-Dolby, pre-201 bending of a tape's transfer function used
> to achieve the best signal-to-noise ratio.
With all due respect, this is flat wrong. Limiting was not only used,
but used extensively when recording music with a wide dynamic range.
Note that by "limiting" here we are not only talking about automatic
limiters (designed to prevent ugly overload), but also "tapering" of
the recorded gain by the engineer in order to achieve the same result,
more gracefully.