Jitter Correlation to Audibility

[stv014]

OK, further reduced jitter (5 times lower random + 10 times lower sinusoid):
 
path30jr.flac

 

[esldude]

   
I did not use a 10 minute recording to measure the relative speed of each DAC compared to the ADC. I simply used my sinetest utility (it can be downloaded from the link in my signature) that can measure frequency accurately - using quadratic interpolation on a large FFT with a Gaussian window - on the recorded JTest signals, which were of course much shorter than 10 minutes

.
 
However, I also tried to find out the absolute speed of the ADC itself, and that is more difficult without the right equipment for measuring frequency with a high accuracy. That is why I played and recorded the very long tone in a loopback configuration on the same sound card, and measured its length in real time with the hope that the system clock (which has a speed automatically adjusted using NTP) is more accurate and can be used as a reference, at least over a sufficiently long time.

Okay, I understand now.  Had it mixed up before. 

 

[Don Hills]

There's a jitter discussion on the pro-audio mailing list at the moment. It's mainly focused on studio practices (master external clock versus internal clocking) but a couple of useful links were posted:
 
Audibility of jitter:
http://www.cranesong.com/A_Matter_%20Of_Time_The%20_Audibility_Of_Clock_Jitter.pdf
 
A set of sample files with jitter:
http://www.cranesong.com/jitter_1.html
 
Discussion of studio clocking and jitter effects:
http://www.soundonsound.com/sos/jun10/articles/masterclocks.htm

 

[jcx]

...These are caused by the Ultracurve’s clock recovery circuitry trying to maintain a lock on the external reference clock signal, and although relatively low in level (most are still below -130dBFS), these spurious signals can have audible effects because they are not directly related to the input signal and are inherently unmusical. So even though a very good-quality external word clock is being supplied here, the performance of the A-D converter becomes noticeably (and audibly) worse than when running on its internal clock...

 I'd like to see the peer reviewed evidence for -130 dB audibility of any pattern of spurious signal of any degree of correlation or "inherently unmusical" character
 
with the condition that those "most are still below -130dBFS" are a consequence of having a -4 dBfs musical signal at reasonable listening level at the same time

 

[nick_charles]

  There's a jitter discussion on the pro-audio mailing list at the moment. It's mainly focused on studio practices (master external clock versus internal clocking) but a couple of useful links were posted:
 
Audibility of jitter:
http://www.cranesong.com/A_Matter_%20Of_Time_The%20_Audibility_Of_Clock_Jitter.pdf
 
A set of sample files with jitter:
http://www.cranesong.com/jitter_1.html
 
Discussion of studio clocking and jitter effects:
http://www.soundonsound.com/sos/jun10/articles/masterclocks.htm

 
The jitter samples are useful so that people can do DBTs if they wish. The rest is standard fare even including a re-education segment where you are corrected if you mistake the jittered file as superior. It is the same old dog and pony show (a minimum of 24db gain so you can hear the difference file !) and calling it an experiment is a shameful use of language. But if UMS can DBT the megajittered sample then that will be interesting !
 
As for audibility of the 6ps jitter signal - I'm a bit skeptical. I ran the samples through audacity and charted them in Excel - there is no difference greater (+/-) than 0.1db between the two at any frequency until the energy is at below -121db

 

[stv014]

As for audibility of the 6ps jitter signal - I'm a bit skeptical. I ran the samples through audacity and charted them in Excel - there is no difference greater (+/-) than 0.1db between the two at any frequency until the energy is at below -121db

 
Well, the hiss that can be heard in the 6 ps difference file seems to be equivalent to ~21-bit TPDF dithered quantization noise (assuming 72 dB gain), so that should give an idea of how much the added jitter matters.
 
By the way, at least for the "800 ps" file, the difference seems to be too large (I did not check the other files). I do not know exactly how this artificial jitter was created, though.

 

[Digitalchkn]

  Discussion of studio clocking and jitter effects:
http://www.soundonsound.com/sos/jun10/articles/masterclocks.htm

I would like to point out that the comparison of the DFT cable vs the mic cable in the Interface-Induced Jitter side-bar is a bit misleading. The description implies that the effect is capacitive, but what it actually points to is transmission line effect. The plateau in the waveform is purely a reflective effect associated with mismatched termination to the cabling at the source of the clock and/or the destination. It would only make sense that a mic cable would not be designed to 75ohm impedance.
 
Needless to say that even though the clock frequency may be 3MHz, but the signal edge rates are contain substantially higher harmonic content (probably well into 100's of MHz in this example) that would clearly result in typical length of cable to appear as a transmission line rather than a lumped capacitance.

 

[esldude]

   
Well, the hiss that can be heard in the 6 ps difference file seems to be equivalent to ~21-bit TPDF dithered quantization noise (assuming 72 dB gain), so that should give an idea of how much the added jitter matters.
 
By the way, at least for the "800 ps" file, the difference seems to be too large (I did not check the other files). I do not know exactly how this artificial jitter was created, though.

 
 
Yes, I checked that too against file B, and it does seem much too large a difference to be from jitter of 800 ps. 

 

[nick_charles]

   
Well, the hiss that can be heard in the 6 ps difference file seems to be equivalent to ~21-bit TPDF dithered quantization noise (assuming 72 dB gain), so that should give an idea of how much the added jitter matters.
 
By the way, at least for the "800 ps" file, the difference seems to be too large (I did not check the other files). I do not know exactly how this artificial jitter was created, though.

 
I tried to use Audio Diffmaker to generate an independent difference file but it kept crashing, I could not work out how to create a difference track from 2 stereo tracks in Audacity, but the difference track provided did indeed seem too much notwithstanding cranking it up by 24db. I've found smaller overall measurable differences comparing files with or without 100ns jitter ! I had a few samples I saved from elsewhere and there was no difference of note (> 0.1db) between the 100ns sample and 0(sic) ns sample until you hit 20K and have a signal of -106db !

 

[UltMusicSnob]

   
I tried to use Audio Diffmaker to generate an independent difference file but it kept crashing, I could not work out how to create a difference track from 2 stereo tracks in Audacity, but the difference track provided did indeed seem too much notwithstanding cranking it up by 24db. I've found smaller overall measurable differences comparing files with or without 100ns jitter ! I had a few samples I saved from elsewhere and there was no difference of note (> 0.1db) between the 100ns sample and 0(sic) ns sample until you hit 20K and have a signal of -106db !

Very quick and dirty test, cheap earphones, from the Crane Song website, this is the reference file B versus the A version.
 
foo_abx 1.3.4 report
foobar2000 v1.2.8
2013/09/26 13:52:13
File A: C:\Users\jhughes1\Downloads\Crane_B.wav
File B: C:\Users\jhughes1\Downloads\Crane_A.wav
13:52:13 : Test started.
13:52:51 : 00/01  100.0%
13:53:31 : 01/02  75.0%
13:53:44 : 02/03  50.0%
13:54:15 : 03/04  31.3%
13:54:26 : 03/05  50.0%
13:54:31 : 04/06  34.4%
13:54:42 : 04/07  50.0%
13:54:48 : 05/08  36.3%
13:54:53 : 06/09  25.4%
13:55:00 : 07/10  17.2%
13:55:14 : 08/11  11.3%
13:55:22 : 09/12  7.3%
13:55:33 : 10/13  4.6%
13:55:55 : 11/14  2.9%
13:56:18 : 12/15  1.8%
13:56:22 : Test finished.
 ----------
Total: 12/15 (1.8%)
 
Listen for bass focus. I was learning as I went, so I would expect to be able to replicate this with a higher reliability in fewer rounds.

 

[stv014]

  I tried to use Audio Diffmaker to generate an independent difference file but it kept crashing, I could not work out how to create a difference track from 2 stereo tracks in Audacity, but the difference track provided did indeed seem too much notwithstanding cranking it up by 24db.

 
There seems to be a small amount of delay (~2 us) and level mismatch (~0.01 dB) between B.wav and E.wav, but so far I was only able to get a difference signal with about equal RMS level to "10hz difference file.wav" without the 24 dB boost. E.wav probably does have much higher jitter than 800 ps.
 
Edit: simply mixing B.wav and E.wav without any delay/gain applied gives the original (downloadable) difference file. So, the above information regarding delay and level mismatch is not correct, my measurements were probably inaccurate because of the high amount of jitter.

 

[stv014]

  Very quick and dirty hear, cheap earphones, from the Crane Song website, this is the reference file B versus the A version.

 
Note that all other files are inverted (have a gain of -1) relative to B.wav. So, you need to fix that first before comparing them.

 

[nick_charles]

I inverted A and compared A(6ps)  vs B  . Most of the time the two files are very close but at higher frequencies they diverge drastically as below
 
 

 
This looks a touch iffy !
 
In comparison here is a graph of two files 1 without jitter and one with 100ns jitter for the same high frequencies
 
 

 

[Don Hills]

   
Note that all other files are inverted (have a gain of -1) relative to B.wav. So, you need to fix that first before comparing them.

 
Note that I was careful not to express any opinion about the quality of the test or the expressed opinions in the articles.
I will say that most mixing and (especially) mastering engineers care deeply about the same minutaie as audiophiles. It's not surprising, because they spend their days listening critically to audio quality on good systems.

 

[stv014]

  Note that I was careful not to express any opinion about the quality of the test or the expressed opinions in the articles.
I will say that most mixing and (especially) mastering engineers care deeply about the same minutaie as audiophiles. It's not surprising, because they spend their days listening critically to audio quality on good systems.

 
How is that relevant to my suggestion to UltMusicSnob to fix the inverted file before an ABX test ?