"Arny Krueger" wrote in message
. ..
"R. Stanton" wrote in message
ps.com...
On Nov 9, 7:31 pm, "Arny Krueger" wrote:
"Harry Lavo" wrote in
messagenews:VeCdnYsnVo0rW87YnZ2dnUVZ_vudnZ2d@comca st.com...
So, some people claim they need to get to 192/24 before it is no
longer
possible to hear a difference from analog. Has this been verified
with
a double bind listening test? I doubt it.Agreed.
http://groups.google.com/group/rec.a...2ca804fb14333e
George Massenburg wrote:"What it isn't is what you might call the
[golden-ear pantload name
here] demonstration where this guy sits you down and plays you a
couple of things (could be anything: the levels aren't calibrated and
could be anywhere). [G.E.P.L.] proceeds to switch sounds for you
saying, "O.K., listen to this. RIght, NOW listen to THIS!" (maybe he
actually turns the monitor gain up) "Wow, that's great, huh?" And this
other? HEY, you couldn't possibly like THAT, could you??? I mean,
c'mon, you'd be an IDIOT not to hear the difference...
Any test where you know which piece of gear you're listening to...any
test that's not perfectly blindfolded and well-controlled cannot
possibly be called scientific. As much as I don't like the downsides
of the A-B-C-Hidden Reference it's a very useful discipline to reveal
modest differences.
"The best listening tests demand that you objectify what you hear.
"An example of a useful, forthright listening test is the high-octave
test suggested and implemented by Bob Katz, where he takes a 96/24
file (presumably rich in 20kHz content), and filters it at 20kHz or
so. Then he listens (through exactly the same hardware, and under
exactly the same circumstances, removing conversion, to name one
factor, as a possible variant) to see if he can tell the difference
between the two (filtered and unfiltered) files. Can I be brave here
and tell you the truth? Neither of us have had significant successes
with differentiating between the samples.
Some folks on, I believe, RAP reported that this was the case in a
comparison of live feed vs. 96/24 vs. 192/24.Reporting something that
is a complete illusion is pretty easy. Please see
the collected posts of Harry Lavo, for many examples.
I'm not surprised that cutting off signal above 20KHz has no audible
effect.
Many people are. In fact, an inaudible brick-wall cutoff can be somewhat
lower than 20 KHz. 16 KHz is a common number that is used in perceptual
coders, even when high SQ is the goal.
I did some flatness and distortion tests on a sound card and on a CD
player. They were both ruler flat. The distortion was less then 0.003%
and 0.01%. That level of distortion is inaudible. Is there something
else that could cause commertial CD's to sound so poor?
Commercial CDs that sound like crap, do so because of other steps in the
process. The strongest influences in the SQ of a recording are, and in my
estimated order of importance:
(1) Artistic content - well-written, well-arranged, well-played music
tends to sound better. Since I record artists with a wide range of skill
levels, and in various degrees of being properly rehearsed, I'm very aware
of this.
(2) The acoustical environment where the recording was made, including the
microphones and microphone technique. Since I record in a variety of
acoustical environments, I'm also very aware of this. Really good venues
can make mediocre musicanship sound not that bad.
(3) The mix-down, editing and mastering. This interacts with (1) and (2).
If (1) and (2) are good, then the mixdown is pretty much set levels and
go, the editing is clean up the start and finish, and the mastering is
pretty straight-forward. If (1) and (2) are substandard, then I have to
monitor the mix like a hawk and do lots of adjustements to restore
balance, I may do a lot of editing to conceal artistic flaws, and
mastering takes a lot of shucking and jiiving to get things to sound right
in a variety of playback environments.
I'm wondering if *jitter* could be the cause of commertial CD's
sounding harsh.
I seriously doubt it.
I do suspect that a lot of perceived sonic problems of particularly early
CD players were due to borderline tracking. I've found that there can be a
lot of error concealment going on, and it doesn't sound so much like error
concealment. Rather it sounds like other flaws like harshness and
emptiness. It can produce a general state of listener malease. It can even
upset the pace and timing of the music.
In the past 25 years there have been two definate and undeniable
improvements in optical disc players - they cost tremendously less money
for a high level of performance, and they can properly track a far wider
range of imperfect discs.
Perhaps in the mass production process of stamping out
of CD's, excessive jitter distortion is introduced.
The data signal that comes into the CD player from the optical pickup is
often very jittery. The buffering and clocking circuits in any CD player
minimize this to an inaudible level, as a matter of course. Not to say
that every CD player ever made always did this right.
While most of what Arny says here is correct (and it is nice to be able to
agree with him for a change) there is one specific statement that addresses
your hypothesis that I take issue with:
"Commercial CDs that sound like crap, do so because of other steps in the
process."
Arny doesn't provide a basis for his conclusion, but a few years ago there
was a long discussion on Usenet (I believe RAP but I am not certain of that)
different CD's from different plants sounding different, and the role of
the producer and/or engineer in giving final approval of the process. The
general thrust was that the differences were likely the degree with which
error correction was required, and the degree to which on-disk jitter needed
to be corrected. There was also some discussion that the problem was worse
when plants were being provided with analog tape and doing their own
production mastering (thus making the analog and digital quality of their
own D/A's an issue). As Arny says, this was perceived to have been more of
a problem in '80's / early '90's than at the time of the discussion.