1

Topic: The second jitter test is online.

The link below is to the sample clips for the second jitter test.

http://hddaudio.net/wordpress/wp-conten … mple-1.m4a

http://hddaudio.net/wordpress/wp-conten … mple-2.m4a

http://hddaudio.net/wordpress/wp-conten … mple-3.m4a

http://hddaudio.net/wordpress/wp-conten … mple-4.m4a

http://hddaudio.net/wordpress/wp-conten … mple-5.m4a

The samples are 0ns, 3ns, 10ns, 30ns, and 100ns.

2

Re: The second jitter test is online.

Again, this jitter is contrived because it is not at all like what you would get in a real system.  It is random jitter, not correlated with the music data, or it's effect on the electrical system of the components.

I also noticed a lot of white-noise in the last sample set, which I know was not in the original recording, so I suspect that the A/D- D/A or DSP software is introducing a lot of noise.

May the listening begin.

Steve N.
Empirical Audio

Last edited by audioengr (2009-04-27 17:55:32)

3

Re: The second jitter test is online.

So you can't hear it?

There are only 10 types of people in the world: Those who understand binary, and those who do not.

4

Re: The second jitter test is online.

I never said that.  I have not started listening to the second tracks yet.

See the original jitter test results here:
http://hddaudio.net/punbb/viewtopic.php?id=4&p=2

I believe I'm the ONLY one that heard any difference in the 4 tracks.  This shows how unresolving the average system is.

Steve N.
Empirical Audio

Last edited by audioengr (2009-04-28 01:00:56)

5

Re: The second jitter test is online.

Hi, Steve.

What is the reference system for identifying the tracks?

6

Re: The second jitter test is online.

Toshiba XP laptop playing Foobar 0.8.3 (unmapped) -> 2 meter Locus-Design Axis USB cable -> Overdrive USB DAC -> 15-foot Balanced Empirical interconnects -> modded JC-1 monoblocks -> Empirical Clarity7 6-foot speaker cables -> Custom 4-foot BG ribbon speakers and 10" Vifa woofer bass-boxes, custom crossovers.  Power cables to the JC-1's are Empirical Top Fuel.

Acoustic treatments are 2 ASC 1/4 round reflecting tube-traps on either side of the wide-screen TV, and absorbing corner pillows.  Side-wall treatments of two Sonex 2X4 foot panels.  Room is 25 foot square with sloping two story ceiling.  Glass along the entire back wall.  Listening position about 10 feet from the speakers, just off-center in the room.  See this partial picture:
http://www.empiricalaudio.com/about-empirical-audio/

I may also try battery power on the Overdrive logic supply or a Pace-Car reclocker driving the I2S on the Overdrive, but I dont think I'll need it.

5 tracks is a lot to parse-out BTW.

Steve N.
Empirical Audio

Last edited by audioengr (2009-04-28 00:58:34)

7

Re: The second jitter test is online.

OK. i have my picks, where are the answers? if i hold my computer upside down, will it be at the bottom of the page?. I am not too confident but I hear a bit o difference, as I did on the first set. Though I only could pick the best of that batch though.

8

Re: The second jitter test is online.

Ed

Which do you feel are best or worse?

9

Re: The second jitter test is online.

audioengr wrote:

Again, this jitter is contrived because it is not at all like what you would get in a real system.  It is random jitter, not correlated with the music data, or it's effect on the electrical system of the components.

This is incorrect.
The jitter spectrum used in the second set of files is very close to that which you would obtain from a typical SPDIF receiver chip - so very much like a real system !  It is true that you can sometimes get periodic components in the jitter (manifest as discrete sidebands in the jitter spectrum), but these are usually at a very low level compared to the noise-like jitter (unless the equipment has been incredibly badly designed).

Furthemore, other similar tests (eg. from the BBC) have shown that the audibility of noise-like and periodic jitter to be very similar, with the noise-like jitter typically being slightly more noticable, when its bandwidth is comparable to the audio itself.

audioengr wrote:

I also noticed a lot of white-noise in the last sample set, which I know was not in the original recording, so I suspect that the A/D- D/A or DSP software is introducing a lot of noise.

The jittering was performed using professional digital signal processing software, not normally available to the amateur (because of its price !).  The noise floor due to the processing when the jitter is set to zero is around 250dB below full scale.

(One of the files in the latest set of 5 has zero jitter).

Chris.

10

Re: The second jitter test is online.

quote audioengr - "I believe I'm the ONLY one that heard any difference in the 4 tracks.  This shows how unresolving the average system is." - unquote

Er, it also could be that you imagined you could hear it. (for whatever reason).

JCBrum.

11

Re: The second jitter test is online.

ABX Test Completed: 2009-04-28 07:30:33 -0400

Number of tests performed: 10
Number of correct answers: 4
Percentage correct:  40%

File 1 = sample-1.m4a
File 2 = sample-3.m4a


n    A    X    B    Choice    Score
1    1    2    2      B         1/1
2    1    2    2      B         2/2
3    1    2    2      A         2/3
4    1    2    2      B         3/4
5    1    2    2      A         3/5
6    1    1    2      B         3/6
7    1    2    2      A         3/7
8    1    2    2      A         3/8
9    1    2    2      A         3/9
10    1    2    2      B         4/10

--------------------------------------------------------------


ABX Test Completed: 2009-04-28 07:32:52 -0400

Number of tests performed: 10
Number of correct answers: 3
Percentage correct:  30%

File 1 = sample-1.m4a
File 2 = sample-4.m4a


n    A    X    B    Choice    Score
1    2    2    1      A         1/1
2    2    2    1      B         1/2
3    2    1    1      A         1/3
4    2    2    1      B         1/4
5    2    2    1      B         1/5
6    2    2    1      B         1/6
7    2    2    1      B         1/7
8    2    2    1      A         2/8
9    2    2    1      B         2/9
10    2    2    1      A         3/10

--------------------------------------------------------------

ABX Test Completed: 2009-04-28 07:32:52 -0400

Number of tests performed: 10
Number of correct answers: 3
Percentage correct:  30%

File 1 = sample-1.m4a
File 2 = sample-4.m4a


n    A    X    B    Choice    Score
1    2    2    1      A         1/1
2    2    2    1      B         1/2
3    2    1    1      A         1/3
4    2    2    1      B         1/4
5    2    2    1      B         1/5
6    2    2    1      B         1/6
7    2    2    1      B         1/7
8    2    2    1      A         2/8
9    2    2    1      B         2/9
10    2    2    1      A         3/10

--------------------------------------------------------------

ABX Test Completed: 2009-04-28 07:35:16 -0400

Number of tests performed: 10
Number of correct answers: 4
Percentage correct:  40%

File 1 = sample-2.m4a
File 2 = sample-5.m4a


n    A    X    B    Choice    Score
1    1    2    2      A         0/1
2    1    2    2      A         0/2
3    1    1    2      B         0/3
4    1    2    2      B         1/4
5    1    1    2      B         1/5
6    1    1    2      A         2/6
7    1    1    2      A         3/7
8    1    2    2      B         4/8
9    1    1    2      B         4/9
10    1    1    2      B         4/10

--------------------------------------------------------------

ABX Test Completed: 2009-04-28 07:41:11 -0400

Number of tests performed: 10
Number of correct answers: 5
Percentage correct:  50%

File 1 = sample-1.m4a
File 2 = sample-5.m4a


n    A    X    B    Choice    Score
1    2    1    1      A         0/1
2    2    1    1      B         1/2
3    2    2    1      B         1/3
4    2    1    1      B         2/4
5    2    1    1      B         3/5
6    2    1    1      A         3/6
7    2    2    1      A         4/7
8    2    2    1      A         5/8
9    2    2    1      B         5/9
10    2    2    1      B         5/10

--------------------------------------------------------------

12

Re: The second jitter test is online.

One more ...

--------------------------------------------------------------


ABX Test Completed: 2009-04-28 07:59:13 -0400

Number of tests performed: 10
Number of correct answers: 5
Percentage correct:  50%

File 1 = sample-4.m4a
File 2 = sample-5.m4a


n    A    X    B    Choice    Score
1    2    2    1      A         1/1
2    2    1    1      A         1/2
3    2    2    1      A         2/3
4    2    2    1      A         3/4
5    2    2    1      A         4/5
6    2    2    1      B         4/6
7    2    1    1      B         5/7
8    2    2    1      B         5/8
9    2    1    1      A         5/9
10    2    2    1      B         5/10

--------------------------------------------------------------

13

Re: The second jitter test is online.

JCBrum wrote:

quote audioengr - "I believe I'm the ONLY one that heard any difference in the 4 tracks.  This shows how unresolving the average system is." - unquote

Er, it also could be that you imagined you could hear it. (for whatever reason).

JCBrum.


I also see fairies sometimes.  They speak to me and help me with the determination.;)

Steve N.

14

Re: The second jitter test is online.

Darren - why M4A files this time?  Why not just use .wav?

I cannot seem to get my Foobar 0.8.3 to play these.

Steve N.

Last edited by audioengr (2009-04-30 01:57:33)

15

Re: The second jitter test is online.

So, if you can't play them Steve, weren't you just a little bit tempted to look at them on your digital scope ? maybe ? wink

JCBrum.

16

Re: The second jitter test is online.

Steve,

I used lossless as its quicker to download. You could convert with dbpoweramp on Windows.

17

Re: The second jitter test is online.

cs wrote:

The jittering was performed using professional digital signal processing software, not normally available to the amateur (because of its price !).  The noise floor due to the processing when the jitter is set to zero is around 250dB below full scale.

Chris.

So where did all of the white-noise come from?  I have this CD and the rips from it and it does not have all of this hash in it.  This is separate from the jitter.

Steve N.

18

Re: The second jitter test is online.

Darren wrote:

Steve,

I used lossless as its quicker to download. You could convert with dbpoweramp on Windows.


Okay, thanks.  I need some experience with dB poweramp anyway.

Steve N.

19

Re: The second jitter test is online.

JCBrum wrote:

So, if you can't play them Steve, weren't you just a little bit tempted to look at them on your digital scope ? maybe ? wink

JCBrum.


Nope.  In fact, a digital scope would show you very little unless the jitter is quite high.  Even one with 2nsec risetime spec and 10GHz sample-rate.  This is a constantly changing waveform.  Seeing the difference in the analog signal is almost impossible.  You could see a difference in the digital signal, but the jitter would have to be high (I dont know what it is in these samples), and you would have to capture hundreds of samples and overlay the same exact time interval and then measure the extremes on particular edges.  You might be able to pick out the best and worst, but not the ones in-between.

This is exactly why this is such a good experiment.  It is difficult to cheat.

Anyway, I would appreciate it if you discontinued accusing me of cheating.  If you are a gentleman, you will stop.

Steve N.

Last edited by audioengr (2009-05-01 00:06:55)

20

Re: The second jitter test is online.

Sorry Steve, but I don't think either you or your claims for your products have any credibilty.  I'm very suspicious of anything that you say.

Having said that I shall not prolong the agony.  No doubt other people will make up their own minds according to their own experiences and preferences.

Regards, JCBrum.

21

Re: The second jitter test is online.

My wife and I took a listen to the 5 new tracks using my Overdrive DAC with Foobar 0.8.3 and 24/96 upsampling.  The differences are just as apparent as on the first tracks, maybe a bit more obvious.  No white-noise to listen-through this time, thanks.  Because there is both a vocalist and a guitar, we only needed the first 5 seconds to pick them out.

One thing I did do was play my own EAC ripped .wav version of the same track to compare this.  It was significantly better than any of the jitter tracks downloaded and converted to .wav files.  I think the DSP or compressed-file conversion of the CODEC to .wav changed the files adversely.  This makes things doubly difficult when the CODEC/compression conversions are messing up the files just as badly as the jitter.

Darren - what tool/version was used to do the original rip and was it on PC or Mac?

We plan to repeat the test this evening, this time with different random numbering and at 44.1, primarily because there were two tracks that seemed very close and we want to nail it this time if possible.

Steve N.
Empirical Audio

Last edited by audioengr (2009-05-01 17:49:24)

22

Re: The second jitter test is online.

JCBrum wrote:

Sorry Steve, but I don't think either you or your claims for your products have any credibilty.  I'm very suspicious of anything that you say.

Having said that I shall not prolong the agony.  No doubt other people will make up their own minds according to their own experiences and preferences.

Regards, JCBrum.


Well Charles Hanson agrees with me about the difficulty of measuring this jitter:
http://www.audioasylum.com/forums/pcaud … 51034.html

You know who he is dont you?

Steve N.

Last edited by audioengr (2009-05-03 18:16:34)

23

Re: The second jitter test is online.

So when you're making these non-ABX 'tests',
which power cable of yours are you using,
and how might they alter your results?

$999 Silver 6ft
$899 Gold 6ft
$1,300 Silver 9ft
$1,200 Gold 9ft

24

Re: The second jitter test is online.

Quote Nugent - "Well Charles Hanson agrees with me about the difficulty of measuring this jitter:  You know who he is dont you? " -  endquote.

Is he the singer ?  or the baby-sitter ?  or did you mean Charles Hansen ?

(hint, this is a rhetorical question.) tongue

JC.

(edit I put a smiley in in case you thought I was having a go at you) big_smile

Last edited by JCBrum (2009-05-05 07:26:43)

25

Re: The second jitter test is online.

audioengr wrote:

I think the DSP or compressed-file conversion of the CODEC to .wav changed the files adversely.  This makes things doubly difficult when the CODEC/compression conversions are messing up the files just as badly as the jitter.

Darren - what tool/version was used to do the original rip and was it on PC or Mac?

I'm sure I answered this before, but anyway.....

The conversion from wav to ALAC and back again was done on iTunes for the Mac.  This is known to be bit perfect, and has been verified many times by many people.  It is a totally deterministic and error free process.

The software used for the jitter processing was Agilent SystemVue, which is professional digital signal processing software (run on a PC).  The error produced by the processing when the jitter is set to zero is around 250dB below fullscale.

Chris.