To Arny,

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation? I
only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.


T

"Thomas A" wrote in message
om

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation?

First positive report I've ever heard of.

I only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Your playback system may have nonlinear distortion that intermodates content
19 KHz down into lower frequencies.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.

Try a better playback system (especially your audio interface on the Mac)
and see what happens.

On 25 Jul 2004 15:51:26 -0700, (Thomas A)
wrote:

To Arny,

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation? I
only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.


T

If you are monitoring your progress as you go (able to see whether
each trial is correct or not), then the ability to stop whenever p
dips below 0.05 constitutes a form of "cherry picking." As the number
of trials increases in such a test (termed a sequential test), the
probability of being able to cherry pick at p 0.05 increases
greatly.

If you fixed the number of trials prior to starting, and then
continued until you reached this pre-determined n, then your test is
(statistically) valid.

ff123

Thomas A wrote:
To Arny,

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation? I
only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.

You sure you can hear 18KHz - not intermod products ? You must be very
young.....


geoff

"Arny Krueger" wrote in message ...
"Thomas A" wrote in message
om

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation?

First positive report I've ever heard of.

How about the 16/32? I started with this test to figure out what to
listen for. How are other people hearing this?

I only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Your playback system may have nonlinear distortion that intermodates content
19 KHz down into lower frequencies.

Yes, agreed.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.

Try a better playback system (especially your audio interface on the Mac)
and see what happens.

I am going to buy the M-audio transit, and try it out. It measures ok,
I guess:

http://www.fixup.net/products/benchm...ransit1644.htm

"Thomas A" wrote in message
om
"Arny Krueger" wrote in message
...
"Thomas A" wrote in message
om

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation?

First positive report I've ever heard of.

How about the 16/32? I started with this test to figure out what to
listen for. How are other people hearing this?

They hear it as sounding the same as the reference.

I only use cheap headphones, but there is a slight noticable
difference in the attack rate of the triangle. However, I can only
faintly hear 18 kHz, nothing at 19 kHz for static tones.

Your playback system may have nonlinear distortion that intermodates
content
19 KHz down into lower frequencies.

Yes, agreed.

In days of yore, a sound card called the PAS-16 that did this.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.

Try a better playback system (especially your audio interface on the
Mac) and see what happens.

I am going to buy the M-audio transit, and try it out. It measures ok,
I guess:

http://www.fixup.net/products/benchm...ransit1644.htm

Yes, that looks pretty good. The problem with USB 1.1 has always been a
tendency towards clicks and pops on some machines.

ff123 wrote in message . ..
On 25 Jul 2004 15:51:26 -0700, (Thomas A)
wrote:

To Arny,

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation? I
only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.


T

If you are monitoring your progress as you go (able to see whether
each trial is correct or not), then the ability to stop whenever p
dips below 0.05 constitutes a form of "cherry picking." As the number
of trials increases in such a test (termed a sequential test), the
probability of being able to cherry pick at p 0.05 increases
greatly.

If you fixed the number of trials prior to starting, and then
continued until you reached this pre-determined n, then your test is
(statistically) valid.

ff123

Determined n=100. One trial, after I going from 16/32 excercise. But
it still can be intermod. distorsion.

T

"Geoff Wood" -nospam wrote in message ...
Thomas A wrote:
To Arny,

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation? I
only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.

You sure you can hear 18KHz - not intermod products ? You must be very
young.....


geoff

That is from my speakers and test CD, not from computer. Depends of
course of SPL, don't remember now what I used. 17 kHz is fine. 18 kHz
is well, very faint. 19 kHz is nothing. I am not very young, but that
is a relative term.

"Arny Krueger" wrote in message ...
"Thomas A" wrote in message
om

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation?

First positive report I've ever heard of.

I only use cheap headphones, but there is a slight noticable difference
in the attack rate of the triangle. However, I can only faintly hear
18 kHz, nothing at 19 kHz for static tones.

Your playback system may have nonlinear distortion that intermodates content
19 KHz down into lower frequencies.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.

Try a better playback system (especially your audio interface on the Mac)
and see what happens.

I remade a test today quite sloppy (I was in a hurry doing other
things) with better headphones using the 16/32 vs 16/44.1 and decided
for n=50. I got 35/50, p0.05. What has people reported with 16/32?

When I get the USB soundcard later this week I will give another
report. Probably the audible phenomenon dissappears.

T

"Thomas A" wrote in message
om
"Arny Krueger" wrote in message
...
"Thomas A" wrote in message
om

I get significant p-values (0.05, n=100) when I test the samples
16/44.1 vs 16/38 downsampled triangle. Is that a common observation?

First positive report I've ever heard of.

I only use cheap headphones, but there is a slight noticable
difference in the attack rate of the triangle. However, I can only
faintly hear 18 kHz, nothing at 19 kHz for static tones.

Your playback system may have nonlinear distortion that intermodates
content
19 KHz down into lower frequencies.

Or perhaps there is some malfunction of the ABX software for mac? I
did not expect to get this value, but I will repeat this again.

Try a better playback system (especially your audio interface on the
Mac) and see what happens.

I remade a test today quite sloppy (I was in a hurry doing other
things) with better headphones using the 16/32 vs 16/44.1 and decided
for n=50. I got 35/50, p0.05. What has people reported with 16/32?

Again, very little. There's a reason why the 28 and 22 KHz SR samples are
provided. ;-)

When I get the USB soundcard later this week I will give another
report. Probably the audible phenomenon dissappears.

Wating on the proverbial pins and needles.