[Audio Empire]

I have seen arguments here that 16-bit, 44.1 KHz "CD" audio is perfect an=
d=20
that there is no difference between it and higher bit-depth and higher=20
sampling frequency recordings or even SACD. While I never actually agreed=
=20
with this opinion, I never disputed it either because I had not been priv=
y to=20
any bias-free listening experiments which would allow me to have a defini=
tive=20
and informed opinion. Well, yesterday, that all changed. I was able to ta=
ke=20
part in a double-blind listening session with several audiophile friends =
that=20
has proven to me beyond any doubt, that high-resolution recordings are=20
definitely worth the effort.=20

Two identical Korg MR-1s were connected in parallel to the same mixer and=
the=20
same pair of good quality condenser microphones (actually it was a single=
=20
point stereo mike, if you want to nit-pick). The recording was of a colle=
ge=20
jazz band and one of the Korgs was set to record at 24-bit, 192 KHz and t=
he=20
other was set to record at 16-bit 44.1 Khz. Both were WAV files.=20

For playback, both Korgs were connected to two line-level inputs of a Kre=
ll=20
KAV-300iL integrated amplifier (200 Watts RMS/channel) feeding a pair of=20
Martin-Logan Vantage speakers (34=AD23,000 Hz =B13dB). The outputs of the=
two=20
Korg recorders were matched exactly using an HP 3400A Audio RMS Voltmeter=
..=20

The Krell amp was switched via the Krell remote control by someone who wa=
s=20
standing BEHIND the listeners and thus did not take part in the evaluatio=
n=20
and did not know which Korg was connected to which input, or which one he=
ld=20
the high-res recording and which one held the standard CD resolution=20
recording. All he was required to do was to note whether we were listenin=
g=20
non-balanced input #1 or non-balanced input #2 on the Krell at any given=20
time.=20

The results were unanimous. The high-res version just sounded better. It =
was=20
easiest to hear the difference on brushed cymbals and brushed snare. They=
=20
just sounded more alive, cleaner, more lifelike and airier. Since I was t=
he=20
one who made the recordings, I noticed that the 24-bit, 192KHz recording=20
sounded much more like what I heard in the rehearsal hall where the=20
recordings were made than did the 16-bit, 44.1 KHz version.=20

There is no doubt. The high-resolution recordings are better than CD, a r=
eal=20
but fairly small step forward, but still, a step forward, nonetheless.=20

[bob]

Sounds like a one-trial DBT to me. Next time, try something with a
shred of statistical significance.

And, as a follow-up, try down-converting the hi-res version to
16/44.1, and comparing it to each of the original recordings.

bob

[Steven Sullivan]

Audio Empire wrote:
I have seen arguments here that 16-bit, 44.1 KHz "CD" audio is perfect

No, you haven't. I defy you to quote anyone who wrote that Redbook audio
(or any audio reproduction for that matter) is *perfect*.

There is no doubt. The high-resolution recordings are better than CD, a real
but fairly small step forward, but still, a step forward, nonetheless.

I see you didn't provide statistics.

--
-S
We have it in our power to begin the world over again - Thomas Paine

[Audio Empire]

On Tue, 4 May 2010 15:45:48 -0700, Dick Pierce wrote
(in article ):

Audio Empire wrote:
I have seen arguments here that 16-bit, 44.1 KHz "CD" audio is perfect

Where?

Well, yesterday, that all changed. I was able to take=20
part in a double-blind listening session with several audiophile friend=
s that=20
has proven to me beyond any doubt, that high-resolution recordings are=20
definitely worth the effort.=20

And your decision is based on some rather shaky assumptions

Two identical Korg MR-1s were connected in parallel to the same mixer a=
nd the=20
same pair of good quality condenser microphones (actually it was a sing=
le=20
point stereo mike, if you want to nit-pick). The recording was of a col=
lege=20
jazz band and one of the Korgs was set to record at 24-bit, 192 KHz and=
the=20
other was set to record at 16-bit 44.1 Khz. Both were WAV files.=20

Here is probably the most important assumption you have
made: that the in-band performance of these two machines
at different sample rate is identical. I see no reason
why that assumption is a priori true, none whatsoever.

So, what in fact you MAY have heard is the difference
in implementation between the two sample rates or
other differences that are NOT instrinsic simply due
to the difference in sample rates and resolution.

For example, what if the anti-aliasing filter algorithm
or other internal processing actually has some very
substantial artifacts? Even if it's the same algorith,
the fact that you have a lousy filter design at 20 kHs
vs the same lousy filter design at 50 kHz is MORE than
enough to account for the difference.

What if, for example, there's substantial artifacts
going an octave below the cutoff frequency? One of them
is going to have artifacts down to 10 kHz, the other
down to 25 kHz. That soprt of difference, alone, has
the potential for making a substantial audible difference,
again having NOTHING to do with the intrinsic difference
between the "resolution."

Another example: what if there is a fixed anti-imaging
filter designed to work properly at the higher sample
rate, and it allows artifacts through to downstream
processing from the lower sample rate that it adequately
takes care of at the higher sample rate (since the
imaging artifacts are at a higher frequency anyway)?

Yet another: How, in fact, does the unit handle different
sample rates? Does it, in fact, change the sample clock
or does it simply decimate the higher rate to get the
lower sampled stream? Do we know that it does it properly?
How do we know that?

You have SO many variables and SO many hidden assumptions
of this type that I think, objectively, the results are
sufficiently unreliable as to be worthless.

Now, having been, in fact, involved in the design and
implementation of A/D and D/A systems for high-quality
editing worksations, I have actually had the opportunity
to carefully evaluate a lot of this kind of equipment.
The number of implementations which were just plain awful
are depressingly common. And sample rate changes and
conversions, even in this modern era, is one area that
seems to trouble a lot of designers.

These are just one realm of many where the implementation
could have problems. There are many others.

So, I ask again: how do you know your REALLY hearing the
difference in resolution and NOT the difference in equipment
performance due to design or implementation?

Now, I just happened to look up Korg's specification for the
MR-1. First, I had forgotten that, as far as digital recording
system are concerned, it's really what I would consider to
be at the lower end of the market.

But the specifications are, frankly, pretty awful in some
respects. Their specification for S/N, for example, is
pretty awful, being only 90 dB A-weighted. I was doing
converter system with real 94+ dB S/N UNWEIGHTED at 16 bits
12-14 years ago.

And, as I suspected, the A/D converters run at a fixed rate,
and the conversion to PCM is done internally by software.
So, again, show me that those software algorithms got it
right. Show me, for example, that the noise-shaping of
the sample rate converters has no audible difference at
different sample rates.

Sorry to say that while your conclusion may seem so obvious,
the data is highly suspect.

There MAY be a real audible difference between 44.1 kHz
and 192 kHz samples rates WHEN PROPERLY AND TRANSPARENTLY
IMPLEMENTED, but a Korg MR-1, not that many steps above
the toy level, is NOT the way to prove it.

For playback, both Korgs were connected to two line-level inputs of a K=
rell=20
KAV-300iL integrated amplifier (200 Watts RMS/channel) feeding a pair o=
f=20
Martin-Logan Vantage speakers (34=AD23,000 Hz =B13dB).

WOW! 3.423 MHz! Now THAT'S a speaker! :-)

I never wrote that, I wrote "34 - 23000 Hz +/- 3dB",; That's "Thirty-four
Herz to twenty-three thousand Herz, plus or minus three deciBels". I have no
control over what the 'rec.audio.high-end' server or your news reader does to
the ASCI text I sent.

"There MAY be a real audible difference between 44.1 kHz
and 192 kHz samples rates WHEN PROPERLY AND TRANSPARENTLY
IMPLEMENTED..."

But of course, that never can be determined. Because no matter what one does,
people who's audio belief system DEPENDS upon high-resolution audio being no
better than CD will always find something wrong with the procedure. That's
fine. I'm fairly convinced and I'm the only one that I need to convince.

As to the Korg being "not that many steps above a toy", apparently, Robert
Woods of Telarc fame disagrees with you. He used an MR-1 to do microphone
placements tests for the Cincinnati and Atlanta orchestras (and others) and
thought the quality of the recordings good enough to release. also, you
might want to look at the graphs on an MR-1 at:

http://www.sonicstudios.com/mr-1revw.htm

You'll see that most of the noise is supersonic and that which isn't is
extremely narrow-band. Yes, the rack-mounted MR-2 at $2K is better, but I
don't want to haul anything that big around (or else I'd just keep hauling
around my Apogee Rosetta 200).

Also if you were talking about a Zoom H2 or H4n as being close to toys, I
would agree with you. They ARE just above a toy (but an H2 at 24/96 makes a
mighty handy back-up recorder, nonetheless). But at almost $900 list, this
thing is hardly "not that many steps above a toy". The recordings that I've
made with it are stupendous sounding. Much better than anything you can buy
commercially and easily the equal of anything I've made with my Apogee
Rosetta 200 and my Mac iBook computer via Audacity or Logic Studio.

[bob]

On May 4, 9:20=A0pm, Audio Empire wrote:

But of course, that never can be determined. Because no matter what one d=
oes,
people who's audio belief system DEPENDS upon high-resolution audio being=
no
better than CD will always find something wrong with the procedure.

Do the test as well as Meyer and Moran did theirs, and I'll believe
you.

What will it take for you to believe me?

I'm fairly convinced and I'm the only one that I need to convince.

I guess, unlike me, you've made up your mind.

bob

[Audio Empire]

On Wed, 5 May 2010 07:14:58 -0700, Dick Pierce wrote
(in article ):

Audio Empire wrote:
On Tue, 4 May 2010 15:45:48 -0700, Dick Pierce wrote
(in article ):

[ Excessive quotation snipped. -- dsr ]


Also if you were talking about a Zoom H2 or H4n as being close to toys, I
would agree with you. They ARE just above a toy (but an H2 at 24/96 makes a
mighty handy back-up recorder, nonetheless). But at almost $900 list, this
thing is hardly "not that many steps above a toy". The recordings that I've
made with it are stupendous sounding. Much better than anything you can buy
commercially and easily the equal of anything I've made with my Apogee
Rosetta 200 and my Mac iBook computer via Audacity or Logic Studio.

Both you and I can buy a LOT "commercially" including the
likes of Apogee and such. This is your claim?



Please go back and re-read what I wrote. When I said " Much better than
anything you can buy commercially..." I was referring to commercial
recordings, not equipment.

[Audio Empire]

On Tue, 4 May 2010 19:20:55 -0700, bob wrote
(in article ):

On May 4, 9:20=A0pm, Audio Empire wrote:

But of course, that never can be determined. Because no matter what one d=
oes,
people who's audio belief system DEPENDS upon high-resolution audio being=
no
better than CD will always find something wrong with the procedure.

Do the test as well as Meyer and Moran did theirs, and I'll believe
you.

You don't seem to understand, Bob. I don't really care whether or not you or
anyone else believe me. I didn't make my OP to convince anyone of anything. I
posted to say that whereas I didn't have a real opinion before, I feel that I
have done my due diligence in this matter, and I do have one now

What will it take for you to believe me?

I'm fairly convinced and I'm the only one that I need to convince.

I guess, unlike me, you've made up your mind.

I plan to do more tests, but, for the here and now, I'm fairly certain that
there is a noticeable difference between 16/44.1 and 24/192 and that the
latter sounds more like the real thing. I haven't yet tried it with 24/96, to
see if one can notice that too.

I do know one thing. I shall continue to record at the higher rates mostly
because of the headroom that 24-bit affords me during recording.

[Arny Krueger]

"Audio Empire" wrote in message
...

The outputs of the two Korg recorders were matched exactly using an HP
3400A Audio RMS Voltmeter.

What sort of signal was used in the level matching phase of your test?

[Audio Empire]

On Wed, 5 May 2010 15:01:15 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message
...

The outputs of the two Korg recorders were matched exactly using an HP
3400A Audio RMS Voltmeter.

What sort of signal was used in the level matching phase of your test?

I recorded a 400 Hz test tone at -6 dB on the Korg's meter at the head of
both recordings. I used that. I don't know the accuracy of the 3400A, but on
the expanded scale the meter was at exactly the same place for each
recording, and it does have a mirror behind the pointer for parallax. I
suspect that since the expanded scale has increments of 0.1 dB, that the
difference was far less than that. Of course, absolute accuracy is not really
important in this case, but relative accuracy is.

[Arny Krueger]

"Audio Empire" wrote in message
...
On Wed, 5 May 2010 15:01:15 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message
...

The outputs of the two Korg recorders were matched exactly using an HP
3400A Audio RMS Voltmeter.

What sort of signal was used in the level matching phase of your test?

I recorded a 400 Hz test tone at -6 dB on the Korg's meter at the head of
both recordings. I used that. I don't know the accuracy of the 3400A, but
on
the expanded scale the meter was at exactly the same place for each
recording, and it does have a mirror behind the pointer for parallax. I
suspect that since the expanded scale has increments of 0.1 dB, that the
difference was far less than that. Of course, absolute accuracy is not
really
important in this case, but relative accuracy is.

It was always our standard to use test tones at 20, 50, 400 or 1 KHz, 10 KHz
and 20 KHz.

We caught a number of surprising problems this way.

We were sure to use analog meters with taut-band movements to avoid problems
with needle stiction.