[net.audio] West of Oz CD

dsj@rabbit.UUCP (David S. Johnson @ ) (10/02/84)

Here are some comments on Tony Lauck's CD versus LP
test on the Sheffield "West of Oz" recording:

(1) First note that the LP is a direct-to-disc recording, so there was
no analog master tape to degrade the sound.  In the September AUDIO,
Bert Whyte (himself the recording engineer on some direct-to-disc
sessions) performed the same comparison, both for this recording,
and for the previous McBroom recording ("Growing up in Hollywood Town").
The "West" CD was made from a digital mastertape (recorded from the
same feed that made the direct-to-disc master), whereas the "Growing up"
CD was made from an analog mastertape.  Whyte thought the "Growing up"
CD was deficient compared to the LP, whereas the "West" CD was
indistinguishable.  He was listening to the LP using high-end equipment.
One might be able to conclude from this that digital CD's are at least
CLOSER to the "realism" of direct-to-disc than are analog mastertapes.

(2) I wonder if Tony took care in matching volumes when doing his test.
This is especially difficult when the test is performed, as he says,
by first listening to the CD all the way through and then putting on
the LP.  Making the problem more difficult is the fact that the CD is
recorded at an exceedingly low volume level (presumable to leave room
for the transients, but I think Sheffield was far to conservative).
In order to play it at a reasonable level, I must set my volume control
at a level FAR above that at which my other CD's blow me out of the room.
(Indeed, at this level the REVOX player's calibration tone representing its
maximum output would require 1000's of watts to reproduce, although there
is no evidence of clipping on the CD and the transients sound great.)
The CD sounds "strikingly better" when played at this volume (room ambience
suddenly becomes audible, etc.), and this could account for the difference
Tony heard.  I would ask Tony to repeat the test, carefully matching volumes
and switching back and forth between the two sources (a double-blind
experiment would be nice, but I'll settle for this).  If there is a
difference, what precisely is it? 

(I'm not going to buy the direct-to-disc version and make the comparison
myself, as the record itself is a disappointment that no amount
of sonic realism can redeem.  It's something like late 60's Judy Collins
- McBroom emphasizes the similarity by closing with a song from
"Who knows where the time goes" - but the singing and song selection
aren't as good, and the arrangements are quite pedestrian.)

David S. Johnson, AT&T Bell Laboratories

lauck@bergil.DEC (10/09/84)

<>
>(1) First note that the LP is a direct-to-disc recording, so there was
>no analog master tape to degrade the sound.  In the September AUDIO,
>Bert Whyte (himself the recording engineer on some direct-to-disc
>sessions) performed the same comparison, both for this recording,
>and for the previous McBroom recording ("Growing up in Hollywood Town").
>The "West" CD was made from a digital mastertape (recorded from the
>same feed that made the direct-to-disc master), whereas the "Growing up"
>Whyte thought the "Growing up"
>CD was deficient compared to the LP, whereas the "West" CD was
>indistinguishable.  He was listening to the LP using high-end equipment.

I'm really surprised that Bert Whyte had difficulty distinguishing the
LP from the CD. The difference is so great that there is no need to
do double blind testing.  You hear the difference and KNOW it's there.  In
fact, I was shocked at the extent of the difference.  I'll have to
get a copy of the September Audio and read the article.  I'm curious
as to the record playing equipment involved.  I didn't bother to buy the
CD of "Growing up", since it would prove nothing except the limitations
of analog tape recorders, probably not the state of the art recorders at
that.  I certainly didn't feel the need to buy the CD lest the analog
disk wear out....

>(2) I wonder if Tony took care in matching volumes when doing his test.
>This is especially difficult when the test is performed, as he says,
>by first listening to the CD all the way through and then putting on
>the LP.  

I took no care in matching the levels, I just played the music.  The
tonal balance was so different that the concept of level matching makes
no sense.  At what frequency do you match levels?  If measuring power,
on which portions of the program material?  After the inital test I
then played various cuts, one cut at a time, and listened at all combinations
of CD or LP louder, or as close as possible by ear.  In each case, the
LP sounded better, even when it was much too soft or much too loud for
my taste and the CD was set "properly".

I then called up my pro digital friend, E. Brad Meyer, who is an
audio writer and recordist.  I told him that he was going to have to
"reconvert me" back into the pro-digital fold.  He suggested that
perhaps Sheffield should have made their CDs by playing a pressing
of one of their direct disk records into a digital recorder.  I
thought he was joking.  Little did I know...

The prescribed treatment was to borrow a SONY PCM 710 and an ABX
double blind comparator box.  I had to swear to level match within .1db.  I
assured him that I had a tone generator and a suitable meter.

My initial tests consisted of just listening through the PCM and 
comparing it to the Sheffield recordings and other audiophile
recordings, including the Performance Recordings Prokofiev 6th
sonata and the Reference Recording Symphonie Fantastic.  It was
immediately clear that any PCM degradation was minimal, subjectively
at least an order of magnitude less than the difference between the
"West of OZ" CD and LP.  There was no question that the LP and the
LP via the PCM both sounded vastly better than the CD.  Perhaps my
friend Brad had been serious in his recutting suggestion.

I then level matched the PCM to within .05 db on each stereo channel.  To
those who haven't done it before, it is a real pain, unless you have
ten turn pots and fancy meters.  It took me two hours, and caused me nearly
as much pain as when I measured the mass of an electron in college physics. 
(My degree was in mathematics.)  The resulting level setting put
the PCM in a poor position, since without an attenuator on the output I
couldn't get a level match with a recording level above -24db.  This
meant I was running effectively at 13 bits.  When the stylus was lifted,
the PCM noise (dither) was clearly audible, making identification of
the PCM immediate.  However, this noise was not detectable when a
record was playing since it was far enough below the record noise of
all the records I played.

The final result: I correctly identified the PCM in 24 out of 42 trials, 
which is not statistically significant.  Subjectively, I felt that any 
differences were at the very limit of my perception.  Perhaps I could
keep up this performance every night for several months and reach
statistic significance, but why?  It became clear to me that the
problem with bad sounding CDs is not in the system itself, since
the CDs sound much worse than the LPs via PCM. (The 710 uses the same encoding 
format as CD's.)

The obvious explaination, one argued publically by many, is that 
recording engineers and record producers haven't learned to use the
new, more accurate digital medium.  I don't buy this.  The technology
has been available for five years.  By now these problems should have
been overcome.  What's going on here?

I hooked the PCM710 up to my video recorder and made some live
recordings of my wife's Steinway piano and my son's 12 string guitar, using
a pair of AGC C451 cardiod condenser microphones.  The result:  the piano 
sounded much like CD piano recordings, not too bad.  The guitar suffered
from moderately severe "digitalis".   Do I need new microphones?

Here's a subject to ponder:  why do digitally mastered LPs sound bad when
digitally copied analog LPs sound good?  Why should analog then digital sound
good while digital then analog sound bad?  Is it a system interaction 
problem?  Is it a psychoacoustic problem?
    

                          Tony Lauck
                             ...decvax!decwrl!rhea!bergil!lauck

rzdz@fluke.UUCP (10/11/84)

> 	 I hooked the PCM710 up to my video recorder and made some live
> 	 recordings of my wife's Steinway piano and my son's 12 string
> 	 guitar, using a pair of AGC C451 cardiod condenser microphones.
> 	 The result:  the piano sounded much like CD piano recordings, not
> 	 too bad.  The guitar suffered from moderately severe "digitalis".
> 	 Do I need new microphones?

I think that you ought to try some different microphones. The C451's have
a tipped response curve (check the curve that AKG includes with them). Their
response is very *level*, but there is perhaps a 4 dB rise between 20 and
20khz. You can hear it in them, as they tend to sound *very* bright. They're
*very* good microphones, they just sound bigger (brighter?) than life.

As a recording engineer, I like to use the 451 for acoustic guitar, and on
mushy piano's (not your Steinway, I hope). The brightness helps emphasis the
attack of the guitar, and gives the piano a brightness that is hard to obtain
with an equalizer.

On the negative side, this excess brightness might be part of what you call
"digitalis". I suggest that you try a pair of the new B&K recording
microphones (4004, 4006 i think) or the Neumann KM84 or KM83. The 84 is
quite flat, and doesn't have the tipped response that the 451 does. They
have a very neutral sound, sometimes almost warm. The 83 (omni) has an
on-axis rise in its response. You can play with its positioning to use this
rise to advantage, or run the mic at 90 degrees incidence to make it flat.

Rick Chinn
John Fluke Mfg. Co MS 232E
PO Box C9090 Everett WA 98206

ihnp4!uw-beaver----\
decvax!microsof     \
ucbvax!lbl-csam      \ 
                      +====!fluke!rzdz
sun                  /
sb1!allegra         /
ssc-vax------------/ 

(206) 356-5232

rmd@hpfcla.UUCP (rmd) (10/12/84)

The tests  which  Tony  Lauck  ran  Using a Sony PCM  adapter  to record
Sheffield  LP's were very  interesting  to me.  This is  something I had
always  wanted  to do  myself  to rebut  anti-digital  arguments.  Being
pro-digital,  I am not surprised at his results which  indicated  that a
digital recording of an LP sounds very much like the LP.

Tony was still  bothered by the fact the LPs do sound  different  and he
couldn't  understand  why Bert  White  in  Audio  didn't  hear the  same
differences  he did.  I believe that these things are largely  explained
by some fairly obvious and common attributes analog systems.

First, it is the nature of digital  processes  to be  repeatable  in the
common  objective  measures  of  frequency   response,  speed  accuracy,
distortion,   etc.  Some  people   argue  that  digital   systems   vary
subjectively  and  that may or may not be true.  However,  no one that I
know of questions the consistency of frequency response, etc.

On the  other  hand,  it is the  nature  of  analog  systems  NOT  to be
repeatable.  It would be possible to acheive  greater  repeatability  by
spending more money in the design and manufacture of equipment, but most
"high-end"  manufacturers  design for  subjective  rather than objective
performance.

There  are in  fact  large  (relative  to  digital)  variations  in  the
objective    performance   of    turntable/arm/cartridge/head-amp/preamp
combinations.  There are also large  variations in the equipment used to
record and cut LP's.  Frequency  response  variations  can easily exceed
5db when 1db is clearly  detectable.  Longterm  playback  speed can vary
over a 1-2% range.

While many would claim that these variances are not important,  they are
certainly  important for CD-LP  comparisons.  As far as I can tell, most
CD-LP  comparisons  are  done  with no good  data  about  the  objective
performance  of the analog chain  involved.  Without such data, there is
no reason to assume that  differences  are due to any reasons other than
the obvious ones.

Rick Dow
hpfcla!rmd

bytebug@pertec.UUCP (roger long) (10/14/84)

One thing to ponder is just what are we using as a reference point?  If
we compare a CD to an LP in a test such as this, and they sound 
different, we are supposed to immediately come to the conclusion that
the LP sounds "right", and the CD is "wrong?"  I've listened to the
"West of Oz" CD on my system, and thought it sounded just fine.

On the other hand, we all heard what the "official" Sheffield line on
digital was before they brought out their line of CDs.  Perhaps they've
mixed them to sound different than their LPs just to fuel the anti-CD
arguments?

I no longer own a turntable, so perhaps I am biased.  But if anyone
wants to bring Ms. McBroom over to my place, I'd be glad to participate
in a double-blind test of CD vs. the real thing. :-)  To make any other
sort of comparison is like comparing apples and oranges.
-- 
	roger long
	pertec computer corp
	{ucbvax!unisoft | scgvaxd | trwrb | felix}!pertec!bytebug

muller@inmet.UUCP (10/16/84)

***
Nice experiment, trying the blind comparison.  A few comments, questions, and
suggestions:  (1) You say analog to digital sounds good, but digital to analog
sounds bad.  Then you ask why.  Umm...seems to me that this is a bit of a
generalization.  As I try to answer this to myself, and as I ponder the results
of your PCM test, I find lots of room for any specimen to be poorly done, but
no reason to explain why A to D is ALWAYS better than D to A.
(2) With regards your piano and guitar recordings, why not try the same setup,
recording the same stuff into a GOOD analog tape machine?  THEN do a comparison
of the two recordings PCM -> VCR vs. Analog tape.  You will probably hear some
differences, but the audible features you describe as characteristic of digital
may be present to some extent in your analog recording.  What might they be?
  Well, maybe what you heard is exactly what your mics heard.  A look at most
mics' spec sheets shows definitely non-flat frequency response.  They were
designed (evolved?) to sound good, not necessarily have flat-looking response.
Perhaps they compensate for other features of the recording medium, features
which digital systems don't quite duplicate?  (Can we assume there was no real
loss of information in the VCR, such as lost bits that were only poorly "fixed"
on playback?  (I don't know much about how VCR's work.))
(3) Have you, or anyone else, dont this same PCM -> VCR test using pink noise
and an analyzer?  Yeah, I know, you can't really expect .5 dB precision from
this, but it sounds like we are talking about more than this anyway.
...
   Looking objectively at your results, you have shown that the digital process
itself is not really making things sound bad.  So where then is the problem?
It must elsewhere in the recording process, either the total composite 
frequency response, or the dynamic range, or non-linearities or so
Thus it would be good to try the other recordings I suggested, just as another
control.

lauck@bergil.DEC (10/17/84)

>I no longer own a turntable, so perhaps I am biased.  But if anyone
>wants to bring Ms. McBroom over to my place, I'd be glad to participate
>in a double-blind test of CD vs. the real thing. :-)  To make any other
>sort of comparison is like comparing apples and oranges.


The test should be three way, CD vs. LP vs. the real thing.  To 
make it fair it would have to be more than blind. :-)

Ms. McBroom on LP sounded more like a real singer than Ms. McBroom on
CD.  The differences involve qualities like air, depth, dimensionality, 
etc.  However, the record is NOT a solo album.  There are other musical 
instruments on the recording, e.g. clarinet.  I am quite familiar with the 
live sound of these instruments and find their LP rendition more accurate.

Incidentally, I didn't think the sound on the CD was bad at all, UNTIL
I heard the LP.  Perhaps you should have kept your turntable.

		Tony Lauck
			...decvax!decwrl!rhea!bergil!lauck