Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › An interesting paper on jitter audibility
New Posts  All Forums:Forum Nav:

An interesting paper on jitter audibility

post #1 of 112
12/17/06 at 1:17pm
Thread Starter 
http://www.jstage.jst.go.jp/article/ast/26/1/50/_pdf

The study provides evidence that audible limits on some forms of jitter, they used a random jitter pattern, may actually be somewhat higher than even the Benjamin and Gannon study which indicated a threshold of about 20ns. In the study above no subjects detected jitter of 250ns and only 25% could detect jitter at 500ns.
Reply
post #2 of 112
12/17/06 at 7:28pm
Major flaws in this study include:

1) systems provided by listener, which included sometimes speakers and headphones. The quality, imaging and resolution of these systems is in question.

2) the randomly modulated jitter - it did not say anything about the spectra of the jitter used. It may be that 60Hz jitter is detectable at 30psec, but not 20kHz jitter.

3) the amplitude of the signal that is being jittered. This will have a large impact. If the most audible jitter is at 60Hz and the music or sounds are at 1kHz, then the jitter may be masked by the louder sounds.

4) were these trained listeners with hearing tested?

Seems to me that a much more controlled test with issues 1-4 addressed using a single high-resolution system (with good room-acoustics), not headphones, and trained listeners would yield a much more accurate conclusion. I know for a fact that the jitter in my own sources is much less than 250psec and I can still hear differences when I improve on this.

Even better, play a variety of music with the reference system using a source with known larger jitter, say 250 psec and then change to a source with know lower jitter, say 50 psec. I would predict that most trained listeners would hear the difference.

Steve N.
Empirical Audio
Manufacturer
Reply
post #3 of 112
12/17/06 at 7:48pm
Is curious on how always someone try to boost "a myth", or try to prove something, that is used as an excuse for something for years, there is always a "but", and all what we hear is that the methodology used was flawed because of this and that. The same happened with the cables and the test Edwood conducted a couple of years ago, nobody accepted the results as according to them the methodology was "flawed"... My concern is why not trying to find what they did right, and accept the results, instead??? Any methodology will have limitations and will be flawed, AFAIK...There is absolutely no way of considering all scenarios while conducting a given test, while I agree that ones are better than others...

Why not trying to prove the opposite that will be far easier as logic indicates, if this is so hard to prove instead, via a flawless methodology if you can find any??? Why not then implementing one yourselves with a better approach, and try to prove the opposite???....Food for thought...
Reply
post #4 of 112
12/17/06 at 8:02pm
Quote:
Originally Posted by Sovkiller View Post
Is curious on how always someone try to boost "a myth", or try to prove something, that is used as an excuse for something for years, there is always a "but", and all what we hear is that the methodology used was flawed because of this and that. The same happened with the cables and the test Edwood conducted a couple of years ago, nobody accepted the results as according to them the methodology was "flawed"... My concern is why not trying to find what they did right, and accept the results, instead??? Any methodology will have limitations and will be flawed, AFAIK...There is absolutely no way of considering all scenarios while conducting a given test, while I agree that ones are better than others...

Why not trying to prove the opposite that will be far easier as logic indicates, if this is so hard to prove instead, via a flawless methodology if you can find any??? Why not then implementing one yourselves with a better approach, and try to prove the opposite???....Food for thought...
The result of 500nsec jitter being audible to only 25% of listeners is not consistent with my experience. That's why. Even at CES, under those contrived conditions I get 100%.

Steve N.
Reply
post #5 of 112
12/17/06 at 8:14pm
Quote:
Originally Posted by audioengr View Post
The result of 500nsec jitter being audible to only 25% of listeners is not consistent with my experience. That's why. Even at CES, under those contrived conditions I get 100%.

Steve N.
Sorry and in this case you are probably right, and in this particualr case you may get better results using a different approach, but what was curious to me, was that this is the same common behaviour that we have seen lately here...Everytime while someone try to prove, or try to conduct a test to prove something, for good or bad, all what we get is observations of how flawed the methodology was.... it seems that the believers are sometimes more skeptics than the actual skeptics...
Reply
post #6 of 112
12/17/06 at 8:25pm
Quote:
Originally Posted by audioengr View Post
Major flaws in this study include:

1) systems provided by listener, which included sometimes speakers and headphones. The quality, imaging and resolution of these systems is in question.

2) the randomly modulated jitter - it did not say anything about the spectra of the jitter used. It may be that 60Hz jitter is detectable at 30psec, but not 20kHz jitter.

3) the amplitude of the signal that is being jittered. This will have a large impact. If the most audible jitter is at 60Hz and the music or sounds are at 1kHz, then the jitter may be masked by the louder sounds.

4) were these trained listeners with hearing tested?
From the article:

Quote:
A total of 23 audio professionals or semi-professionals participated as the listeners. They were audio engineers, audio critics, sound engineers, and musicians.
So, these listeners are not trained enough? When you do this test at CES, do you only give it to trained listeners? If so, how long is the training, and who does the training? You? Do you really expect us to view your opinion with the same integrity as these academicians? They have very little to gain or lose from this study, while you, someone who sells cables and tweaks...I think we know who butters your bread. Sorry, but if you want to compete with scientists, you need to show us some hard data. You can't just tell anecdotes, and expect us to be convinced. Not those of us with Ph.D.'s, at least.
Reply
post #7 of 112
12/17/06 at 8:46pm
To be fair audioengr is rather biased in that he makes a living on digital gear. That said... 20ns is a BIG number, 500ns is worse than the worst case transport by a lot.

Back when the Monarchy DIP was created we were plagued with 2ns transports, currently if I ball parked it it's closer to 500ps. People knit pick on coax vs. optical over the ~20ps it adds in jitter so I have a hard time believing that people could not hear 500ns or 500,000ps of jitter.

But if its true, why in the world are we wasting our time getting it down any further instead of working on other areas?
Reply
post #8 of 112
12/17/06 at 8:59pm
Thread Starter 
Quote:
Originally Posted by Solude View Post
To be fair audioengr is rather biased in that he makes a living on digital gear. That said... 20ns is a BIG number, 500ns is worse than the worst case transport by a lot.

Back when the Monarchy DIP was created we were plagued with 2ns transports, currently if I ball parked it it's closer to 500ps. People knit pick on coax vs. optical over the ~20ps it adds in jitter so I have a hard time believing that people could not hear 500ns or 500,000ps of jitter.

But if its true, why in the world are we wasting our time getting it down any further instead of working on other areas?
500ns may seem like a big number but it is only half a millisecond i.e 0.0005 seconds.

In the paper they cite studies where jitter in excess of 1 ms is inaudible but without seeing the cited studies it is impossible to judge hiw good they are.
Reply
post #9 of 112
12/17/06 at 9:05pm
Quote:
Originally Posted by hciman77 View Post
500ns may seem like a big number but it is only half a millisecond i.e 0.0005 seconds.
You need to go back to the textbooks, dude:

1 ms (milliseconds) = 10^3 us (microseconds) = 10^6 ns (nanoseconds) = 10^9 ps (picoseconds) = 12^12 fs (femtoseconds)

500 ns is one half of one microsecond.
Reply
post #10 of 112
12/17/06 at 10:24pm
Quote:
Originally Posted by ezkcdude View Post
From the article:



So, these listeners are not trained enough? When you do this test at CES, do you only give it to trained listeners? If so, how long is the training, and who does the training? You? Do you really expect us to view your opinion with the same integrity as these academicians? They have very little to gain or lose from this study, while you, someone who sells cables and tweaks...I think we know who butters your bread. Sorry, but if you want to compete with scientists, you need to show us some hard data. You can't just tell anecdotes, and expect us to be convinced. Not those of us with Ph.D.'s, at least.
If the feedback that I've gotten from "professional" sound engineers is any indication, then no, I dont consider them "trained" listeners. Most audiophiles are better trained IMO. It is their obsession.

As for hard data, I consider this paper anecdotal. My listening tests at CES are also anecdotal. When it comes to correlating effects like jitter to audibility there are too many factors that are overlooked in the tests that I have read about. Same with correlating sound of different cables. There will always be non-believers and believers.

As for convincing you, the only way that I could do this is to demonstrate it to you and you convince yourself. If you were coming to CES, I would be more than happy to. I believe that nothing that I could write here would convince you.

Steve N.
Reply
post #11 of 112
12/17/06 at 10:31pm
Quote:
Originally Posted by Solude View Post
To be fair audioengr is rather biased in that he makes a living on digital gear. That said... 20ns is a BIG number, 500ns is worse than the worst case transport by a lot.

Back when the Monarchy DIP was created we were plagued with 2ns transports, currently if I ball parked it it's closer to 500ps. People knit pick on coax vs. optical over the ~20ps it adds in jitter so I have a hard time believing that people could not hear 500ns or 500,000ps of jitter.

But if its true, why in the world are we wasting our time getting it down any further instead of working on other areas?
I for one, I am "wasting my time" on jitter because:

1) I believe it is the one thing that ultimately keeps digital audio from challenging vinyl/analog

2) I have heard the holy grail, namely inaudible jitter, only for a minute or so, but long to realize that it is truly a revelation. You have no idea how jitter mucks things up until it is completely gone.

Steve N.
Reply
post #12 of 112
12/17/06 at 10:46pm
What I do believe is that the small differences in jitter that exist nowadays among the different decent digital sources that you can find in the market, are small enough, and good enough to satisfy the best listeners, and you will not hear a **** of the difference among them. Same as the differences between two well done and very good cables, your brain will be fooling you all the time, and unless in an strictly controlled DBT in which someone can demonstrate me that is able to point them and succeed 100% of the times, I would hesitate to believe either one...

I prefer and I will keep on enjoying my music regardless of how much jitter I would hear, at the end there is no such a thing as a perfect source anyway, all of them offer pros and contras...and I like the contras of the digital better than the contras of the analog ones I have heard...
Reply
post #13 of 112
12/18/06 at 2:18am
Thread Starter 
Quote:
Originally Posted by audioengr View Post
3) the amplitude of the signal that is being jittered. This will have a large impact. If the most audible jitter is at 60Hz and the music or sounds are at 1kHz, then the jitter may be masked by the louder sounds.
Models of jitter audibility indicate that jitter is theoretically most "audible" at higher frequencies i.e 11 - 20khz where the masking effects should be lower. Arny Kreuger's site shows the sidebands from adding 60 hz jitter at varying levels to a 11khz signal.

http://www.pcabx.com/technical/jitter_power/

He includes samples of jittered signals, the -80db is equivalent to the jitter figures from a poor PCDP with anti-shock enabled, this is absolutely inaudible to me under blind testing. However the one home CD player that he tests for jitter shows a - 130db on the digital output and - 105 analog

http://www.pcavtech.com/play-rec/cd6...dex.htm#JIT_AD

this is the Marantz cd67se which has legendarily high jitter figures - rated at over 500ps by TNT.
Reply
post #14 of 112
12/18/06 at 4:30am
I dunno if it's because of the jitter but i do hear a difference in cheap versus expensive cdplayers. is it because of the better analogue output stages or is it because of better jitter control, dunno, it just sounds alot better!
Reply
post #15 of 112
12/18/06 at 5:19am
Quote:
Originally Posted by audioengr View Post
Major flaws in this study include:

1) systems provided by listener, which included sometimes speakers and headphones. The quality, imaging and resolution of these systems is in question.

2) the randomly modulated jitter - it did not say anything about the spectra of the jitter used. It may be that 60Hz jitter is detectable at 30psec, but not 20kHz jitter.

3) the amplitude of the signal that is being jittered. This will have a large impact. If the most audible jitter is at 60Hz and the music or sounds are at 1kHz, then the jitter may be masked by the louder sounds.

4) were these trained listeners with hearing tested?

Seems to me that a much more controlled test with issues 1-4 addressed using a single high-resolution system (with good room-acoustics), not headphones, and trained listeners would yield a much more accurate conclusion. I know for a fact that the jitter in my own sources is much less than 250psec and I can still hear differences when I improve on this.

Even better, play a variety of music with the reference system using a source with known larger jitter, say 250 psec and then change to a source with know lower jitter, say 50 psec. I would predict that most trained listeners would hear the difference.

Steve N.
Empirical Audio
Manufacturer
waio thats a very academic arguement. im speechless.
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Dedicated Source Components
Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › An interesting paper on jitter audibility