[Harry Lavo]

Well, I guess it is a step in the right direction. But it gets us back only
to a point just short of "CD Quality". Now how about a "super-premium" SACD
or DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

[bob]

On Apr 2, 6:06 pm, "Harry Lavo" wrote:
Well, I guess it is a step in the right direction. But it gets us back only
to a point just short of "CD Quality". Now how about a "super-premium" SACD
or DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...emium+music/21...

I suspect the market for even these will be small. 128kbps AAC is
plenty good enough for earbuds and computer speakers. The DRM rules
don't really pinch most consumers that much, unless they want to put
music on non-iPod players, and right now, everybody who'll be buying
these tunes uses iPods. Also, assuming this becomes the industry norm,
you can buy the 99-cent version now, and upgrade for 30 cents later if
you want the better quality or unrestricted use.

One report I read said that, while the price of individual songs will
go up, the album price won't. That's great for us "mature" listeners
who are used to buying music in 40- to 70-minute chunks.

bob

[Steven Sullivan]

Harry Lavo wrote:
Well, I guess it is a step in the right direction. But it gets us back only
to a point just short of "CD Quality". Now how about a "super-premium" SACD
or DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

Again, before complaining about sound quality, have you even tried ABXing (long term, short
term, whatever) a well-made MP3 versus .wav source? It's not hard to set up.

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[MC]

"Harry Lavo" wrote in message
...
Well, I guess it is a step in the right direction. But it gets us back
only
to a point just short of "CD Quality". Now how about a "super-premium"
SACD
or DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498


I think it is a gigantic step in the right direction. It is a chance for
market forces to work against DRM.

Polls indicate that the DRM-free format is going to be outrageously more
popular than the DRMed version.

[Harry Lavo]

"Steven Sullivan" wrote in message
...
Harry Lavo wrote:
Well, I guess it is a step in the right direction. But it gets us back
only
to a point just short of "CD Quality". Now how about a "super-premium"
SACD
or DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

Again, before complaining about sound quality, have you even tried ABXing
(long term, short
term, whatever) a well-made MP3 versus .wav source? It's not hard to set
up.

Where in the above do you see me complaining about sound quality?

[Philip Homburg]

In article ,
Steven Sullivan wrote:
Harry Lavo wrote:
Well, I guess it is a step in the right direction. But it gets us back only
to a point just short of "CD Quality". Now how about a "super-premium" SACD
or DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

Again, before complaining about sound quality, have you even tried ABXing (long term, short
term, whatever) a well-made MP3 versus .wav source? It's not hard to set up.

I just tried 'lame cbr 320', and I could distinguish that (with some
difficulty) from the original using cheap components. I found a suitable spot
in the first track I tried (The first track of 'Amused To Death'), remembered
the differences and got to someting like 17 correct out of 21.

I have no idea whether 'lame' with just a '-b 320' produces 'well-made' MP3s
or not.

--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency

[Arny Krueger]

"Steven Sullivan" wrote in message

Harry Lavo wrote:

Well, I guess it is a step in the right direction. But
it gets us back only to a point just short of "CD
Quality".

Unfortunately two disparate qualities are intermixed - copy protection and
sound quality.

Now how about a "super-premium" SACD or DVD-A
multi-channel" release.

Would they sound any different at all?

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

Again, before complaining about sound quality, have you
even tried ABXing (long term, short term, whatever) a
well-made MP3 versus .wav source? It's not hard to set
up.

In fact, this is one of those cases where PCABX is an exact solution.
However, it is still ABX, scientific, and therefore prone to produce results
that don't agree with some people's personal desires.

[Steven Sullivan]

Philip Homburg wrote:
In article ,
Steven Sullivan wrote:
Harry Lavo wrote:
Well, I guess it is a step in the right direction. But it gets us back only
to a point just short of "CD Quality". Now how about a "super-premium" SACD
or DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

Again, before complaining about sound quality, have you even tried ABXing (long term, short
term, whatever) a well-made MP3 versus .wav source? It's not hard to set up.

I just tried 'lame cbr 320', and I could distinguish that (with some
difficulty) from the original using cheap components.

Using which ABX software?

I found a suitable spot
in the first track I tried (The first track of 'Amused To Death'), remembered
the differences and got to someting like 17 correct out of 21.

I have that disc -- it's one of the few using 'holographic sound'.
What was the suitable spot?

I have no idea whether 'lame' with just a '-b 320' produces 'well-made' MP3s
or not.

It should!

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Harry Lavo]

"Arny Krueger" wrote in message
...
"Steven Sullivan" wrote in message

Harry Lavo wrote:

Well, I guess it is a step in the right direction. But
it gets us back only to a point just short of "CD
Quality".

Unfortunately two disparate qualities are intermixed - copy protection and
sound quality.

Now how about a "super-premium" SACD or DVD-A
multi-channel" release.

Would they sound any different at all?

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

Again, before complaining about sound quality, have you
even tried ABXing (long term, short term, whatever) a
well-made MP3 versus .wav source? It's not hard to set
up.

In fact, this is one of those cases where PCABX is an exact solution.
However, it is still ABX, scientific, and therefore prone to produce
results
that don't agree with some people's personal desires.

I suspect you are talking about me among others. I find the test reported
in response to Steven's post interesting; I have told you that one of the
few times I ever tried to listen to MP3's I leterally couldn't stand it, so
why would I ABX it. In general, I don't fool around with compressed music
at all, so I have little interest in using ABX to explore codecs. I will
concede that ABX is probably well suited for that, since that was the
purpose of its invention. But exploring codecs is not any real interest to
me, nor is computer audio in general except for recording.

[Arny Krueger]

"Philip Homburg" wrote in message

In article ,
Steven Sullivan wrote:
Harry Lavo wrote:
Well, I guess it is a step in the right direction. But
it gets us back only to a point just short of "CD
Quality". Now how about a "super-premium" SACD or
DVD-A multi-channel" release.

http://news.com.com/EMI%2C+Apple+par...ml?tag=nl.e498

Again, before complaining about sound quality, have you
even tried ABXing (long term, short term, whatever) a
well-made MP3 versus .wav source? It's not hard to set
up.

I just tried 'lame cbr 320', and I could distinguish that
(with some difficulty) from the original using cheap
components.

How did you ensure level-matching, time-synchronization and bias controls?

[Arny Krueger]

"Harry Lavo" wrote in message
...
"Arny Krueger" wrote in message
...

In fact, this is one of those cases where PCABX is an exact solution.
However, it is still ABX, scientific, and therefore prone to produce
results that don't agree with some people's personal desires.

I find the test reported
in response to Steven's post interesting; I have told you that one of the
few times I ever tried to listen to MP3's I literally couldn't stand it,
so
why would I ABX it.

You might be surprised, or not.

I've long suspected that some people are very susceptible to bias, and will
report extreme reactions in sighted evaluations of sounds that in fact
differ very little from what they are used to. Prove me wrong! ;-)

In general, I don't fool around with compressed music
at all, so I have little interest in using ABX to explore codecs. I will
concede that ABX is probably well suited for that, since that was the
purpose of its invention.

Simply not true. ABX was invented to compare audio components. ABX was used
long before many people were interested in exploring codecs because back in
1975 codecs just weren't much of an issue for high quality audio.

[Philip Homburg]

In article ,
Arny Krueger wrote:
"Philip Homburg" wrote in message
I just tried 'lame cbr 320', and I could distinguish that
(with some difficulty) from the original using cheap
components.

How did you ensure level-matching, time-synchronization and bias controls?

I use the abx.c program from pcabx.com. The difference mode, even at +20 dB
was mostly quiet during the less complicated parts of the track.

--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency

[Philip Homburg]

In article ,
Steven Sullivan wrote:
Using which ABX software?

I used the abx.c program that can be found on pcabx.com.

I have that disc -- it's one of the few using 'holographic sound'.
What was the suitable spot?

Around 2:15.

--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency

[Steven Sullivan]

Philip Homburg wrote:
In article ,
Steven Sullivan wrote:
Using which ABX software?

I used the abx.c program that can be found on pcabx.com.

I presume the 'click' problem reported for pcabx in the link below been fixed?

http://ff123.net/cymbals.html

Note how his scores fell when he provided his own 'fix' for it.

Also note the scores for 320 kbps -- using LAME, ff123, who is very
well-trained in hearing mp3 artifacts, got 12/16 for the Radiohead sample.
12/16 is the minimum score to achieve for a conclusion of 'likely heard difference' in 16 trials
(p=0.038), and this was from 2001 (LAME 3.87a); LAME has been improved
quite a bit since then (currently 3.97).

So assuming you're using a modern release of LAME,
your *much* better result strikes me as...curious. Is your high
frequency hearing extra-good?

I have that disc -- it's one of the few using 'holographic sound'.
What was the suitable spot?

Around 2:15.

Thanks. I'll test myself via WinABX.

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Harry Lavo]

"Arny Krueger" wrote in message
...
"Harry Lavo" wrote in message
...
"Arny Krueger" wrote in message
...

In fact, this is one of those cases where PCABX is an exact solution.
However, it is still ABX, scientific, and therefore prone to produce
results that don't agree with some people's personal desires.

I find the test reported
in response to Steven's post interesting; I have told you that one of the
few times I ever tried to listen to MP3's I literally couldn't stand it,
so
why would I ABX it.

You might be surprised, or not.

I've long suspected that some people are very susceptible to bias, and
will
report extreme reactions in sighted evaluations of sounds that in fact
differ very little from what they are used to. Prove me wrong! ;-)

I feel no need to prove you wrong, or myself right. ;-\


In general, I don't fool around with compressed music
at all, so I have little interest in using ABX to explore codecs. I will
concede that ABX is probably well suited for that, since that was the
purpose of its invention.

Simply not true. ABX was invented to compare audio components. ABX was
used
long before many people were interested in exploring codecs because back
in
1975 codecs just weren't much of an issue for high quality audio.

But they and related technologies were of interest to the telephone
companies.

[Arny Krueger]

"Philip Homburg" wrote in message

In article ,
Arny Krueger wrote:
"Philip Homburg" wrote in message
I just tried 'lame cbr 320', and I could distinguish
that (with some difficulty) from the original using
cheap components.

When you say distinguish, what were your scores like?

How did you ensure level-matching, time-synchronization
and bias controls?

I use the abx.c program from pcabx.com. The difference
mode, even at +20 dB was mostly quiet during the less
complicated parts of the track.

abx.c doesn't do anything about level matching. AFAIK, none of the ABX do
level-matching.

I've found that some coders aren't exactly unity gain, so they will make the
music louder or software when they encode it.

[Arny Krueger]

"Harry Lavo" wrote in message

"Arny Krueger" wrote in message
...
"Harry Lavo" wrote in message
...
"Arny Krueger" wrote in message
...

In fact, this is one of those cases where PCABX is an
exact solution. However, it is still ABX, scientific,
and therefore prone to produce results that don't
agree with some people's personal desires.

I find the test reported
in response to Steven's post interesting; I have told
you that one of the few times I ever tried to listen to
MP3's I literally couldn't stand it, so
why would I ABX it.

You might be surprised, or not.

I've long suspected that some people are very
susceptible to bias, and will
report extreme reactions in sighted evaluations of
sounds that in fact differ very little from what they
are used to. Prove me wrong! ;-)

I feel no need to prove you wrong, or myself right. ;-\

I feel no need to give much weight to people who don't care about developing
reliable evidence. IOW, talk is cheap.

In general, I don't fool around with compressed music
at all, so I have little interest in using ABX to
explore codecs. I will concede that ABX is probably
well suited for that, since that was the purpose of its
invention.

Simply not true. ABX was invented to compare audio
components. ABX was used
long before many people were interested in exploring
codecs because back in
1975 codecs just weren't much of an issue for high
quality audio.

But they and related technologies were of interest to
the telephone companies.

What's unclear about the phrase "high quality audio", or do you think that
that somehow equates to telephone-grade audio?

The point is Harry that you're now arguing with one of the people who
developed ABX, about what ABX was developed to do.

For the record, speaking as an initial developer of ABX, ABX was developed
first and foremost to compare audio components playing music. The first ABX
test compared two hi fi power amps using a LP as the musical source.

[Steven Sullivan]

Steven Sullivan wrote:
Philip Homburg wrote:
In article ,
Steven Sullivan wrote:
Using which ABX software?

I used the abx.c program that can be found on pcabx.com.

I presume the 'click' problem reported for pcabx in the link below been fixed?

http://ff123.net/cymbals.html

Note how his scores fell when he provided his own 'fix' for it.

Also note the scores for 320 kbps -- using LAME, ff123, who is very
well-trained in hearing mp3 artifacts, got 12/16 for the Radiohead sample.
12/16 is the minimum score to achieve for a conclusion of 'likely heard difference' in 16 trials
(p=0.038), and this was from 2001 (LAME 3.87a); LAME has been improved
quite a bit since then (currently 3.97).

So assuming you're using a modern release of LAME,
your *much* better result strikes me as...curious. Is your high
frequency hearing extra-good?

I have that disc -- it's one of the few using 'holographic sound'.
What was the suitable spot?

Around 2:15.

Thanks. I'll test myself via WinABX.

Tried it, gave up after five trials -- I couldn't 'hear' any difference
even during the A/B switching, meaning I was just guessing (and my guesses turned
out to be random).

(encoded using LAME 3.97, at --insane setting [= 320 kbps CBR], from the 'gold disc' version
of the Waters album, ripped with Exact Audio Copy. I listened to a snippet from 2:14 to 2:19,
using headphones from an M-Audio 2496 soundcard.)

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Steven Sullivan]

In case anyone else wants to test their skill (or luck) at an ABX of a well-made mp3 --

here is a 6-second snippet (well within 'fair use' limits) from 'The Ballad of Bill Hubbard',
the first track of the Roger Waters 'Amused to Death" CD (Sony gold edition), from
approximately 2.12-2.20, in wav (1.4 Mb) and mp3 (LAME 3.97, -b 320, 318 kb) formats

http://www.m-ideas.com/sullivan/mp3/Hubbard_sample.wav
http://www.m-ideas.com/sullivan/mp3/Hubbard_sample.mp3

Adobe Audition statistics indicate that these differ in avg level and peak level by only a few
hundredths of a dB. The mp3 cuts off steeply at 20 kHz, while the wav has low-level (
-110dB) content out to 22 kHz.

ABX tools:
pcabx:
http://www.pcabx.com/training/license.htm

WinABX
http://www.kikeg.arrakis.es/winabx/

foobar2000 (an highly user-configurable sound file player that includes a WinABX-type
comparator):
http://www.foobar2000.org/

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Harry Lavo]

"Arny Krueger" wrote in message
...
"Harry Lavo" wrote in message

"Arny Krueger" wrote in message
...
"Harry Lavo" wrote in message
...
"Arny Krueger" wrote in message
...

In fact, this is one of those cases where PCABX is an
exact solution. However, it is still ABX, scientific,
and therefore prone to produce results that don't
agree with some people's personal desires.

I find the test reported
in response to Steven's post interesting; I have told
you that one of the few times I ever tried to listen to
MP3's I literally couldn't stand it, so
why would I ABX it.

You might be surprised, or not.

I've long suspected that some people are very
susceptible to bias, and will
report extreme reactions in sighted evaluations of
sounds that in fact differ very little from what they
are used to. Prove me wrong! ;-)

I feel no need to prove you wrong, or myself right. ;-\

I feel no need to give much weight to people who don't care about
developing
reliable evidence. IOW, talk is cheap.

In general, I don't fool around with compressed music
at all, so I have little interest in using ABX to
explore codecs. I will concede that ABX is probably
well suited for that, since that was the purpose of its
invention.

Simply not true. ABX was invented to compare audio
components. ABX was used
long before many people were interested in exploring
codecs because back in
1975 codecs just weren't much of an issue for high
quality audio.

But they and related technologies were of interest to
the telephone companies.

What's unclear about the phrase "high quality audio", or do you think that
that somehow equates to telephone-grade audio?

The point is Harry that you're now arguing with one of the people who
developed ABX, about what ABX was developed to do.

For the record, speaking as an initial developer of ABX, ABX was developed
first and foremost to compare audio components playing music. The first
ABX
test compared two hi fi power amps using a LP as the musical source.

We only have your and your group's word that you were the first. The guys
at CBS labs may have been doing something very similar as early as
1972...what proof do we have the one of you didn't just talk to a friend at
a party, or read some obscure reference to it without even remembering where
you heard it from? You'd be the first to say "isn't it strange that some
amateur audiophile group came up with a test better than the pros in the
field at ATT and CBS labs, wouldn't you?"

[Harry Lavo]

"Steven Sullivan" wrote in message
...
Steven Sullivan wrote:
Philip Homburg wrote:
In article ,
Steven Sullivan wrote:
Using which ABX software?

I used the abx.c program that can be found on pcabx.com.

I presume the 'click' problem reported for pcabx in the link below been
fixed?

http://ff123.net/cymbals.html

Note how his scores fell when he provided his own 'fix' for it.

Also note the scores for 320 kbps -- using LAME, ff123, who is very
well-trained in hearing mp3 artifacts, got 12/16 for the Radiohead
sample.
12/16 is the minimum score to achieve for a conclusion of 'likely heard
difference' in 16 trials
(p=0.038), and this was from 2001 (LAME 3.87a); LAME has been improved
quite a bit since then (currently 3.97).

So assuming you're using a modern release of LAME,
your *much* better result strikes me as...curious. Is your high
frequency hearing extra-good?

I have that disc -- it's one of the few using 'holographic sound'.
What was the suitable spot?

Around 2:15.

Thanks. I'll test myself via WinABX.

Tried it, gave up after five trials -- I couldn't 'hear' any difference
even during the A/B switching, meaning I was just guessing (and my guesses
turned
out to be random).

(encoded using LAME 3.97, at --insane setting [= 320 kbps CBR], from the
'gold disc' version
of the Waters album, ripped with Exact Audio Copy. I listened to a
snippet from 2:14 to 2:19,
using headphones from an M-Audio 2496 soundcard.)

I hope you don't consider this a legitimate test. Rule #1 for ABX testing
or any mulitple trial based testing: you must determine in advance how many
trials are going to be done and not stop or know the answer until you've
done them. Nobody can predict what the outcome will be, negative as well as
positive. What you are supposed to do is not guess, rest, and go back to
the samples again and not choose until and unless you have a basis for
making a choice. Perhaps this is what happens when you go in with a bias
towards not hearing something?

[Harry Lavo]

"Steven Sullivan" wrote in message
...
In case anyone else wants to test their skill (or luck) at an ABX of a
well-made mp3 --

here is a 6-second snippet (well within 'fair use' limits) from 'The
Ballad of Bill Hubbard',
the first track of the Roger Waters 'Amused to Death" CD (Sony gold
edition), from
approximately 2.12-2.20, in wav (1.4 Mb) and mp3 (LAME 3.97, -b 320, 318
kb) formats

http://www.m-ideas.com/sullivan/mp3/Hubbard_sample.wav
http://www.m-ideas.com/sullivan/mp3/Hubbard_sample.mp3

Adobe Audition statistics indicate that these differ in avg level and peak
level by only a few
hundredths of a dB. The mp3 cuts off steeply at 20 kHz, while the wav has
low-level (
-110dB) content out to 22 kHz.

ABX tools:
pcabx:
http://www.pcabx.com/training/license.htm

WinABX
http://www.kikeg.arrakis.es/winabx/

foobar2000 (an highly user-configurable sound file player that includes a
WinABX-type
comparator):
http://www.foobar2000.org/

That's a contribution, Steven. Thanks.

[Steven Sullivan]

Harry Lavo wrote:
"Steven Sullivan" wrote in message
...
Steven Sullivan wrote:
Philip Homburg wrote:
In article ,
Steven Sullivan wrote:
Using which ABX software?

I used the abx.c program that can be found on pcabx.com.

I presume the 'click' problem reported for pcabx in the link below been
fixed?

http://ff123.net/cymbals.html

Note how his scores fell when he provided his own 'fix' for it.

Also note the scores for 320 kbps -- using LAME, ff123, who is very
well-trained in hearing mp3 artifacts, got 12/16 for the Radiohead
sample.
12/16 is the minimum score to achieve for a conclusion of 'likely heard
difference' in 16 trials
(p=0.038), and this was from 2001 (LAME 3.87a); LAME has been improved
quite a bit since then (currently 3.97).

So assuming you're using a modern release of LAME,
your *much* better result strikes me as...curious. Is your high
frequency hearing extra-good?

I have that disc -- it's one of the few using 'holographic sound'.
What was the suitable spot?

Around 2:15.

Thanks. I'll test myself via WinABX.

Tried it, gave up after five trials -- I couldn't 'hear' any difference
even during the A/B switching, meaning I was just guessing (and my guesses
turned
out to be random).

(encoded using LAME 3.97, at --insane setting [= 320 kbps CBR], from the
'gold disc' version
of the Waters album, ripped with Exact Audio Copy. I listened to a
snippet from 2:14 to 2:19,
using headphones from an M-Audio 2496 soundcard.)

I hope you don't consider this a legitimate test.

Indeed it wasn't. Continuing with an ABX where I am literally just guessing, would not produce
a legitimate result evidence for or against the reality of a difference I claimed to hear.

No heard difference = nothing to test!

This is expanded on he

"Corollary : only give answers of which you are absolutely certain ! If you have the slightest
doubt, don't answer anything. Take your time. Make pauses. You can stop the test and go on
another day, but never try to guess by "intuition". If you make some mistakes, you will never
have the occasion to do the test again, because anyone will be able to accuse you of making
numbers tell what you want, by "starting again until it works".

http://www.hydrogenaudio.org/forums/...howtopic=16295

Rule #1 for ABX testing
or any mulitple trial based testing: you must determine in advance how many
trials are going to be done and not stop or know the answer until you've
done them. Nobody can predict what the outcome will be, negative as well as
positive. What you are supposed to do is not guess, rest, and go back to
the samples again and not choose until and unless you have a basis for
making a choice. Perhaps this is what happens when you go in with a bias
towards not hearing something?

Harry, I didn't know what my score was until I decided to stop. I stopped because one
shouldn't resort to 'guessing' during an ABX, and it quickly became apparent to me -- from
*listening* -- that that's all I was doing.

But what the heck, I've gone ahead and done 11 more. As before, I spent some time with each
trial, switching between X, A, and B, using both 'continual play' (where the music is seamless
after switching) and 'go back to start' options for each trial, in attempt to 'hear' a
convincing difference. I also tried different sections of the music. Alas, at no time was I
convinced I heard a difference, so again I was literally just guessing. I got 2/5 on the
first round reported above, and a measly 3/11 on the second round. That's cumulative 5/16 ,
which has a p = 0.962....not very promising of having heard a difference. It is undoubtedly
true however that by this point I am skeptical of hearing difference in most musical samples,
using high bitrates. I'm afriad this result won't help remove that 'bias'.

Philip on the other hand scored hugely in the positive in something like 21 trials; unless
those correct answers all occurred after the first five trials, I'm going to assume he really
believed he 'heard' difference early on: not guessing. This was definitely not the case for
me, with my own 320 kbs encode of the same music he used. I had wondered if perhaps there was
something exceptionally difficult to encode in this music, producing audible artifacts, but
for me, at least, that does not appoear to be the case.

YOU however report that mp3s are all but unlistenable. THus you should do considerably better
than me, in the same test. I've provided the materials -- so take the test yourself. Try not
to guess.

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Arny Krueger]

"Harry Lavo" wrote in message

"Arny Krueger" wrote in message
...

For the record, speaking as an initial developer of ABX,
ABX was developed first and foremost to compare audio
components playing music. The first ABX
test compared two hi fi power amps using a LP as the
musical source.

We only have your and your group's word that you were the
first.

No, you also have the JAES review board's concurrence.

That was an issue that came up during refereeing of Clark's article.

You also have the lack of critical comment on that point after the
publication of the article back in the late 70s.

[Harry Lavo]

"Steven Sullivan" wrote in message
...
Harry Lavo wrote:
"Steven Sullivan" wrote in message
...
Steven Sullivan wrote:
Philip Homburg wrote:
In article ,
Steven Sullivan wrote:
Using which ABX software?

I used the abx.c program that can be found on pcabx.com.

I presume the 'click' problem reported for pcabx in the link below
been
fixed?

http://ff123.net/cymbals.html

Note how his scores fell when he provided his own 'fix' for it.

Also note the scores for 320 kbps -- using LAME, ff123, who is very
well-trained in hearing mp3 artifacts, got 12/16 for the Radiohead
sample.
12/16 is the minimum score to achieve for a conclusion of 'likely
heard
difference' in 16 trials
(p=0.038), and this was from 2001 (LAME 3.87a); LAME has been improved
quite a bit since then (currently 3.97).

So assuming you're using a modern release of LAME,
your *much* better result strikes me as...curious. Is your high
frequency hearing extra-good?

I have that disc -- it's one of the few using 'holographic sound'.
What was the suitable spot?

Around 2:15.

Thanks. I'll test myself via WinABX.

Tried it, gave up after five trials -- I couldn't 'hear' any difference
even during the A/B switching, meaning I was just guessing (and my
guesses
turned
out to be random).

(encoded using LAME 3.97, at --insane setting [= 320 kbps CBR], from
the
'gold disc' version
of the Waters album, ripped with Exact Audio Copy. I listened to a
snippet from 2:14 to 2:19,
using headphones from an M-Audio 2496 soundcard.)

I hope you don't consider this a legitimate test.

Indeed it wasn't. Continuing with an ABX where I am literally just
guessing, would not produce
a legitimate result evidence for or against the reality of a
difference I claimed to hear.

No heard difference = nothing to test!

This is expanded on he

"Corollary : only give answers of which you are absolutely certain ! If
you have the slightest
doubt, don't answer anything. Take your time. Make pauses. You can stop
the test and go on
another day, but never try to guess by "intuition". If you make some
mistakes, you will never
have the occasion to do the test again, because anyone will be able to
accuse you of making
numbers tell what you want, by "starting again until it works".

http://www.hydrogenaudio.org/forums/...howtopic=16295

Rule #1 for ABX testing
or any mulitple trial based testing: you must determine in advance how
many
trials are going to be done and not stop or know the answer until you've
done them. Nobody can predict what the outcome will be, negative as well
as
positive. What you are supposed to do is not guess, rest, and go back to
the samples again and not choose until and unless you have a basis for
making a choice. Perhaps this is what happens when you go in with a bias
towards not hearing something?

Harry, I didn't know what my score was until I decided to stop. I stopped
because one
shouldn't resort to 'guessing' during an ABX, and it quickly became
apparent to me -- from
*listening* -- that that's all I was doing.

But what the heck, I've gone ahead and done 11 more. As before, I spent
some time with each
trial, switching between X, A, and B, using both 'continual play' (where
the music is seamless
after switching) and 'go back to start' options for each trial, in attempt
to 'hear' a
convincing difference. I also tried different sections of the music.
Alas, at no time was I
convinced I heard a difference, so again I was literally just guessing. I
got 2/5 on the
first round reported above, and a measly 3/11 on the second round. That's
cumulative 5/16 ,
which has a p = 0.962....not very promising of having heard a difference.
It is undoubtedly
true however that by this point I am skeptical of hearing difference in
most musical samples,
using high bitrates. I'm afriad this result won't help remove that
'bias'.

Philip on the other hand scored hugely in the positive in something like
21 trials; unless
those correct answers all occurred after the first five trials, I'm going
to assume he really
believed he 'heard' difference early on: not guessing. This was
definitely not the case for
me, with my own 320 kbs encode of the same music he used. I had wondered
if perhaps there was
something exceptionally difficult to encode in this music, producing
audible artifacts, but
for me, at least, that does not appoear to be the case.

YOU however report that mp3s are all but unlistenable. THus you should do
considerably better
than me, in the same test. I've provided the materials -- so take the
test yourself. Try not
to guess.

I didn't say all mp3's are unlistenable...just the one's I burned at 196k in
one of the DAW programs I was using (can't even remember which one, or what
the mp3 plugin was). I am so disinterested in portable music (the only
legitimate reason I can see for mp3 or other compression schemes) that I
simply am not interested in pursuing it further. I think I did it at a time
when Arny (or maybe it was you) was proclaiming the virtues of mp3 and I
just wanted a quick trial for myself. Just so you know, years ago I ripped
probably a dozen and a-half CD's to my laptop using RealAudio and found them
highly listenable on the laptop. But the laptop crashed and I lost it all
and never really had the incentive to do it again. If I did, it would
probably be the same...to copy some CD's onto the laptop at high-bit rate
WME just for the few occassions where I am travelling with the laptop and
get stuck....might listen while playing some MS golf or flying my Cessna
:-).

[Philip Homburg]

In article ,
Arny Krueger wrote:
abx.c doesn't do anything about level matching. AFAIK, none of the ABX do
level-matching.

Please check your facts before posting. The 'Loudness' test in the training
section in pcabx.com completely fails in abx.c because of the level
matching. I had to add an extra option to disable it.

Anyhow, if the difference is zero, no addition tweaking is required.

(Meanwhile I failed two times to duplicate my original result)

--
That was it. Done. The faulty Monk was turned out into the desert where it
could believe what it liked, including the idea that it had been hard done
by. It was allowed to keep its horse, since horses were so cheap to make.
-- Douglas Adams in Dirk Gently's Holistic Detective Agency

[Steven Sullivan]

Harry Lavo wrote:

I didn't say all mp3's are unlistenable...just the one's I burned at 196k in
one of the DAW programs I was using (can't even remember which one, or what
the mp3 plugin was).

I can provide the same sample as a 192 kbps mp3.

I am so disinterested in portable music (the only

Pet peeve : disinterested does not mean uninterested. It means 'unbiased'
or 'neutral'.

legitimate reason I can see for mp3 or other compression schemes) that I
simply am not interested in pursuing it further. I think I did it at a time
when Arny (or maybe it was you) was proclaiming the virtues of mp3 and I
just wanted a quick trial for myself. Just so you know, years ago I ripped
probably a dozen and a-half CD's to my laptop using RealAudio and found them
highly listenable on the laptop.

Hmmm. RealAudio uses its *own* proprietary audio formats.

But the laptop crashed and I lost it all
and never really had the incentive to do it again. If I did, it would
probably be the same...to copy some CD's onto the laptop at high-bit rate
WME just for the few occassions where I am travelling with the laptop and
get stuck....might listen while playing some MS golf or flying my Cessna
:-).

So, to sum up, you've never done an ABX of a well-made mp3 versus source.

___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Arny Krueger]

"Steven Sullivan" wrote in message


YOU however report that mp3s are all but unlistenable.
THus you should do considerably better than me, in the
same test. I've provided the materials -- so take the
test yourself.

I don't know about anybody else, but I generally can reliably detect the
difference between "listenable" and "unlistenable" with a high degree of
statistical confidence.

Try not to guess.

Actually, there's nothing wrong with guessing on ABX tests, since they are
so effective at identifying false positives. I do it all the time. I even
occasionally obtain statistically signficant results that way. Thing is, I
would then be talking about a subtle difference, not the difference between
"listenable" and "unlistenable" .

[Arny Krueger]

"Philip Homburg" wrote in message

In article ,
Arny Krueger wrote:
abx.c doesn't do anything about level matching. AFAIK,
none of the ABX do level-matching.

Please check your facts before posting. The 'Loudness'
test in the training section in pcabx.com completely
fails in abx.c because of the level matching. I had to
add an extra option to disable it.

Thank you for the correction.

(Meanwhile I failed two times to duplicate my original
result)

Interesting. Do keep us informed.