Debate Thread: Scott's Hi-res Audio Test - Page 27 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 741Likes
 
Thread Tools
post #781 of 2920 Old 06-20-2014, 04:43 AM
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,420
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 851 Post(s)
Liked: 1242
Quote:
Originally Posted by Frank Derks View Post
Ouch, sine still at the beginning


Training with A B already made it clear to me I couldn't hear the difference with this file I was hearing in the other files.


Test results with only a couple of trails.


foo_abx 1.3.4 report
foobar2000 v1.3.2
2014/06/20 12:49:13
File A: \\diskstationone\music\jitter\test\30 Hz threshold jitter 0.00312.flac
File B: \\diskstationone\music\jitter\test\no jitter.flac
12:49:13 : Test started.
12:51:57 : 00/01 100.0%
12:52:30 : 00/02 100.0%
12:53:44 : 01/03 87.5%
12:57:05 : 02/04 68.8%
12:58:15 : 03/05 50.0%
12:59:11 : 03/06 65.6%
13:00:14 : 03/07 77.3%
13:00:17 : Test finished.
----------
Total: 3/7 (77.3%)
Looks like random guessing to me!

Thanks for trying.

I think the point has been made.

The spectral analysis of the 1 KHz test tones at the beginning of the file is as follows:



If this test were repeated with the 11 KHz 44.1 J-test tone the peaks would be about 20 dB larger - at -44 dB FS instead of -64 dB FS.

It is safe to say that 30 Hz J-test jitter peaking at -80 dBFS or below is several orders of magnitude below the threshold of hearing.
Attached Thumbnails
Click image for larger version

Name:	30 Hz threshold jitter 0.00312.png
Views:	150
Size:	27.4 KB
ID:	127914  

Last edited by arnyk; 06-20-2014 at 04:46 AM.
arnyk is offline  
Sponsored Links
Advertisement
 
post #782 of 2920 Old 06-20-2014, 05:07 AM
AVS Forum Special Member
 
Frank Derks's Avatar
 
Join Date: Jul 2002
Location: Region A,B,C
Posts: 2,373
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 459 Post(s)
Liked: 335
Thanks for the work on this. It is a great opportunity to test how jitter relates to audible differences and get back to the basics.


It's actually appalling how scientific work and results from the past is dismissed offhand. That work in the past had to be done with far more limited test equipment than what is available today.


It's a bit like the old say 'in that we used to have ships of wood and man of steel. Now we have ships of steel and the man are ...
Frank Derks is offline  
post #783 of 2920 Old 06-20-2014, 05:42 AM
AVS Forum Addicted Member
 
DonH50's Avatar
 
Join Date: Feb 2010
Location: Monument CO
Posts: 10,415
Mentioned: 31 Post(s)
Tagged: 0 Thread(s)
Quoted: 2207 Post(s)
Liked: 1805
I've given up posting any listening results because all my previous experience was on older equipment, and of course my old MG-IIIa's and 50+ year old ears are irrelevant to modern audiophiles. I've auditioned some newer equipment but have not found any that makes it worth the 10x or greater price difference. And yes, depending upon the amp and the speakers, the amps sounded different. Especially when a small tube amp was compared to a larger SS amp. As for differences among SS amps, eh, not so much, but that's obviously my lack of experience, judgement, and hearing acuity.

As for my technical knowledge, thankfully my career does not depend upon defending it in an audio forum on the Internet. Decades-old grad acoustics classes and all the engineering experience is meaningless and out of date. Measurements are irrelevant. Theory is irrelevant. Luckily for me, at my day job the science still seems to apply, so I can happily live in my insular little world. Which is not free from jitter, unfortunately, though at GHz data rates we look at it a little differently.

I see the same things at the trumpet forum I moderate; seasoned pros and top-tier players get shouted down by high schoolers and amateurs (some amateurs are very good, btw, but they are not the ones arguing) with plenty of time and attitude. Sad thing is the top-notch players don't need the hassle and often just quit posting, and everyone loses a valuable resource.

/rant

"After silence, that which best expresses the inexpressible, is music" - Aldous Huxley
DonH50 is offline  
 
post #784 of 2920 Old 06-20-2014, 05:49 AM
Mark Henninger
 
imagic's Avatar
 
Join Date: Dec 2005
Location: Philadelphia, PA
Posts: 12,446
Mentioned: 277 Post(s)
Tagged: 0 Thread(s)
Quoted: 6597 Post(s)
Liked: 11350
Quote:
Originally Posted by koturban View Post
The topic is jitter. Please stay on topic.

Thanks.
Lol. The topic is not jitter. The current argument is about jitter. The topic is "Scott's Hi-res Audio Test."

Mark Henninger, Senior Editor at AVS Forum

Last edited by imagic; 06-20-2014 at 05:52 AM.
imagic is offline  
post #785 of 2920 Old 06-20-2014, 06:14 AM
AVS Forum Special Member
 
Frank Derks's Avatar
 
Join Date: Jul 2002
Location: Region A,B,C
Posts: 2,373
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 459 Post(s)
Liked: 335
The jitter argument is now settled.


We can conclude that the commonly encountered jitter levels in digital audio are not audible with a 16bit 44.1kHz source and it isn't going to be audible with a hirez source either.


I knew that already but it was good to do the test to get a point of reference on it's audibility threshold.
Other forms of (random) jitter might only elevate the noise floor and chances of hearing this are zilch.
Frank Derks is offline  
post #786 of 2920 Old 06-20-2014, 06:19 AM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by arnyk View Post
The bar has been set artificially high. Many of the listening tests that I've participated in are documented here:

http://home.provide.net/~djcarlst/abx_data.htm

Many of the techincal tests that I've done were documented at www.pcabx.com which is still acessible via the well-known wayback machine.

It's not the JAES, but its real.
Good morning Arny. I take anything that is documented. Simply posting the results is of limited value. Let's say I told you that I could tell the difference between 4 CD players and gave you the stats. You would accept that data? Or would you ask for the details of the test condition as to find a problem and hence, invalidate the results?

Quote:
Originally Posted by arnyk View Post
Why would I do more work like this only to have it be dismissed as has been done with my previous work?
I don't know Arny. I was addressing your suggestion: "It strikes me that many of these issues can be resolved pretty quickly with some well-designed listening tests. In the end its all about sound quality, and the most relevant judge of sound quality involves human beings doing the best possible job of listening, right?"

I have highlighted the key part of your previous post. So that low IQ people like me get it, which is it Arny? That the tests can resolve this argument "pretty quickly" or be a waste of time? And why?

Quote:
Originally Posted by arnyk View Post
Actually the web has a goodly number of AVR technical tests on the Sound and Vision web site, as well as Audioholics. Why do I need to reinvent the wheel?
Hmmm. You said that current AVRs are good enough performance wise. We don't even have the measurements of your AVR or that of any other people posting with respect to noise/modulation/jitter. If you do have that data, then my apologies. Please post them.

If you don't have them, then I would think it is inappropriate to draw conclusions with regards to audibility seeing how we have no listening tests either.

I mean either we are data driven or not. You have been using my measurements to make some of your points. What would you do without them?
amirm is offline  
post #787 of 2920 Old 06-20-2014, 06:56 AM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by arnyk View Post
That's for anybody who wants to judge using readily available evidence.
Arny, you created the test. The listeners have no choice but to live with those conditions. I am asking about your intentions in creating this specific test. In this thread you have been saying that the lower the frequency of jitter, the more masking makes it potentially inaudible. In this test, you have picked the 30 Hz frequency. If your logic is correct on masking, which it is , then it reasons that you created a listening test that gave the lowest odds of someone hearing real life jitter. As such it would be improper to draw any conclusions from this one test.

If we really care about audibility of jitter in real products we need to simulate far more frequencies and all at once as the devices output.

Quote:
Originally Posted by arnyk View Post
Check out the FFT analysis of the 1 KHz tone at the beginning of each file. Each whole file was processed the same way.
That is insufficient for proper analysis Arny. Signal processing issues may show up in the music segment but not the 1 KHz tone. Is there a reason you are being secretive and not telling us how the files are created? We ultimately need to know that what we tested is what is purported to be.
amirm is offline  
post #788 of 2920 Old 06-20-2014, 07:07 AM
AVS Forum Special Member
 
Frank Derks's Avatar
 
Join Date: Jul 2002
Location: Region A,B,C
Posts: 2,373
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 459 Post(s)
Liked: 335
This listener is perfectly fine with the conditions.
The 30Hz Jitter at various levels isolates the jitter issue from other possible issues sufficiently to determine the audible threshold.


How the files are created is no secret and is revealed in this thread already.
That is if you understand the science behind this...
Frank Derks is offline  
post #789 of 2920 Old 06-20-2014, 07:10 AM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by krabapple View Post
This HA post in Interpretation of a blind listening test (also by Pio) contains a useful note to keep in mind . Again, the discussion there is mainly in the context of MP3 codec testings, but applies generally --

If we are testing the existence of Superman, and get a positive answer, that is "Superman really exists because the probability of the null hypothesis is less than 5%". Must we accept the existence of Superman ? Is it an infallible, scientific proof of its existence ?

No, it's just chance. Getting an event whose probability is less than 5% is not uncommon.
Gosh. It seems we have an argument for everything. We keep saying listening tests rule but then set the stage to ignore them with incomplete analogies.

Yes, proof of superman existing has a very high bar. Let's look at its inverse. Assume superman does exist but he is never seen in his "superman" outfit. You only see him as an ordinary person. What are the odds of you finding the truth that superman exists? Zero. Yet, I just said he does exist.

That is the problem with a lot of listening tests. They do not start with a revealing test condition. They pick random test files and run them by people. Take Arny's test. Who he wants to guarantee that there is no more revealing music file for his test? I hope no one .

When it comes to small distortions, music selection becomes critical. See this article I wrote on audibility of small distortions.

The reference to MP3 testing is a good one. We don't just pick tracks at random to establish fidelity of an MP3 encoder. We pick carefully selected tracks that have shown to be revealing of compression artifacts. That science is pretty well known. When it comes to types of distortions we are talking about, the nature of device distortions is not well understood due to lack of data. Even if we had such data, there is no analysis given for why the file is revealing of such distortions.

I think there are some 10 million tracks in typical online markets like iTunes. We want our testing to cover many of them. A random sampling is not going to do that unless we can demonstrate that it acts like most of those files.

Let's consider Arny's listening tests. As distortions became smaller, I used specific slices of 2-3 seconds to detect differences. Finding and picking such small segments in a track is hard and prone to mistakes. The proper file would be revealing across most if not all of its track, making it much easier to find the artifacts.
amirm is offline  
post #790 of 2920 Old 06-20-2014, 07:14 AM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by Frank Derks View Post
This listener is perfectly fine with the conditions.
Well, this listener was not .

Quote:
Originally Posted by Frank Derks View Post
The 30Hz Jitter at various levels isolates the jitter issue from other possible issues sufficiently to determine the audible threshold.
No, it is not sufficient for anything remotely like that. Pls see my last post.

Quote:
Originally Posted by Frank Derks View Post
How the files are created is no secret and is revealed in this thread already.
That is if you understand the science behind this...
Well, I am not as smart and knowledgeable as you all . Remember, I found an issue with the first set of files which Arny himself did not know about. "Peer" review is not just for papers. Let's review the actual steps and see if we can duplicate Arny's test results.
amirm is offline  
post #791 of 2920 Old 06-20-2014, 07:21 AM
AVS Forum Special Member
 
Frank Derks's Avatar
 
Join Date: Jul 2002
Location: Region A,B,C
Posts: 2,373
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 459 Post(s)
Liked: 335
Quote:
Originally Posted by amirm View Post
.....

Let's consider Arny's listening tests. As distortions became smaller, I used specific slices of 2-3 seconds to detect differences. Finding and picking such small segments in a track is hard and prone to mistakes. The proper file would be revealing across most if not all of its track, making it much easier to find the artifacts.

Exactly what I did.


A higher resoltion file and/ther music selection *might* shift the audibility threshold in this particular test 6 dB lower but that's about it.
This is still much higher than the encountered jitter levels.
Frank Derks is offline  
post #792 of 2920 Old 06-20-2014, 07:36 AM
AVS Forum Special Member
 
Frank Derks's Avatar
 
Join Date: Jul 2002
Location: Region A,B,C
Posts: 2,373
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 459 Post(s)
Liked: 335
Quote:
Originally Posted by amirm View Post
Well, this listener was not .


No, it is not sufficient for anything remotely like that. Pls see my last post.


Well, I am not as smart and knowledgeable as you all . Remember, I found an issue with the first set of files which Arny himself did not know about. "Peer" review is not just for papers. Let's review the actual steps and see if we can duplicate Arny's test results.
First I didn't see your results from the last two files with the lowest jitter amounts posted yet.


Second I dismiss the ramblings in that post off hand. It's nothing more than that. ramblings that do not carry weight regardless off how well written the post is. Are you expecting that the masking thresholds mysteriously shift to lower levels with jitter at 60Hz? ( Do you expect that Usain Bolt can run the 100 meter under 5 seconds by switching to another brand of running shoes?)


Third. Why not create your own test and create some files we can test and post results about.
Arny asked for source material and jitter levels and frequencies more to your liking. So far nothing suggested.
Frank Derks is offline  
post #793 of 2920 Old 06-20-2014, 08:18 AM
AVS Forum Special Member
 
RobertR's Avatar
 
Join Date: Dec 2000
Location: California
Posts: 6,645
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 494 Post(s)
Liked: 660
Quote:
Originally Posted by Frank Derks View Post
First I didn't see your results from the last two files with the lowest jitter amounts posted yet.


Second I dismiss the ramblings in that post off hand. It's nothing more than that. ramblings that do not carry weight regardless off how well written the post is. Are you expecting that the masking thresholds mysteriously shift to lower levels with jitter at 60Hz? ( Do you expect that Usain Bolt can run the 100 meter under 5 seconds by switching to another brand of running shoes?)


Third. Why not create your own test and create some files we can test and post results about.
That would require him to put up or shut up. Neither is likely to happen.
andyc56 likes this.
RobertR is offline  
post #794 of 2920 Old 06-20-2014, 10:02 AM
 
spkr's Avatar
 
Join Date: Apr 2013
Posts: 816
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 459 Post(s)
Liked: 153
Quote:
Originally Posted by urapnes1 View Post
I don't think what I wrote was whining,
You are free to think whatever you want.
Quote:
Originally Posted by urapnes1 View Post
I signed on to this thread to try to learn more about jitter.
These will help you to set your bearings straight. Link 1 Link 2 Link 3 Link 4 Link 5
imagic likes this.
spkr is offline  
post #795 of 2920 Old 06-20-2014, 10:21 AM
AVS Forum Addicted Member
 
Chu Gai's Avatar
 
Join Date: Sep 2002
Location: NYC area
Posts: 15,247
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 584 Post(s)
Liked: 817
Arny, I think can be reasona ly accommodating, Amir. Can you suggest something that would be more revealing or discriminating wrt jitter?

And Don, don't forget amateurs built the ark but professionals built the Titanic!

"I've found that when you want to know the truth about someone that someone is probably the last person you should ask." - Gregory House
Chu Gai is offline  
post #796 of 2920 Old 06-20-2014, 11:02 AM
AVS Forum Special Member
 
fotto's Avatar
 
Join Date: Nov 2003
Location: N.E. OH
Posts: 1,877
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 21 Post(s)
Liked: 21
Quote:
Originally Posted by spkr View Post
These will help you to set your bearings straight. Link 1 Link 2 Link 3 Link 4 Link 5
Decent summary.
fotto is offline  
post #797 of 2920 Old 06-20-2014, 11:48 AM
AVS Forum Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 17,528
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1487 Post(s)
Liked: 1191
Good afternoon Amir. Hope you had a restful evening.

I still would like to know if the poor audio due to jitter using HDMI is really audible. And yet again... how can one use S/PDIF to achieve the "DTS/DD" hi-rez audio formats?
(you may need to "dumb" it down for some of us. )

I know you are busy with your own forum and well as educating us on AVS, but I would like to get some direction and insight.

Thanks in advance for your answer and continued support on AVS.
andyc56 likes this.
Ratman is offline  
post #798 of 2920 Old 06-20-2014, 11:52 AM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by arnyk View Post
Not audio.
Please forgive me for saying so but that is not remotely correct Arny [that HDMI clock is not used for audio]. The HDMI incoming clock is used to capture all data. This includes the PCM audio samples (or bit stream in the case of compressed formats). Without it, you won't get a picture *or* sound! It is the heart of the system.

That clock then is used to derive the audio DAC clock. That is done because audio must stay in sync with video or you get lip sync problems which is very annoying and easy to notice.

This is all beside the point as you are moving the goal post (I am sure there is a debating term for that ). Here is what TVOD and I were responding to:

Quote:
Originally Posted by arnyk View Post
There is no such thing as a HDMI clock. Check the HDMI interface specs. No clock line, no clock signal, just digital data. Just like S/PDIF and many other digital audio interfaces. The clock signal is intermixed with the data.
Your reference is to *HDMI* clock. You didn't say audio but per above, even if you had it would make no difference.

You talk about HDMI specs. Here is a quote from HDMI CEC spec:

"Audio data being carried across the HDMI link, which is driven by a TMDS clock running at a rate corresponding to the video pixel rate, does not retain the original audio sample clock. The task of recreating this clock at the Sink is called Audio Clock Regeneration."

And if that doesn't make it abundantly clear, here is a recommended architecture for clock extraction for audio (on the right) from HDMI TDMS clock (on the left):



There is also this interesting note on the topic of our discussion:

If there is a constant fractional relationship between these two clocks, and the two clocks are exactly synchronous, then the CTS value will quickly come to a constant value. If the clocks are asynchronous, or there is some amount of jitter between the two clocks, then the CTS value will typically alternate between two or three different values. Greater variations are possible with larger jitter.

Nothing like the spec telling the implementor that having jitter is OK!

Honestly Arny, you are saying that a car has no wheels. Can you please indicate which spec you read on HDMI that said none of is true? I can't imagine there is any document any place that backs your assertion. I think we are past "weak science" and into pseudo science to say what you have Arny.
amirm is offline  
post #799 of 2920 Old 06-20-2014, 12:05 PM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by Ratman View Post
I still would like to know if the poor audio due to jitter using HDMI is really audible.
No one can answer that because we do not remotely have enough sampling of HDMI products to know how they perform. You can ask if your HDMI AVR has audible jitter but then you have to present me with its measurements. If we have that, we can try to apply our understanding of psychoacoustics to figure that out.

Your question is akin to asking me if the fish *you* buy at your grocery store is safe to eat when all you have seen is a sampling of 5 grocery stores elsewhere in the country.

Quote:
Originally Posted by Ratman View Post
And yet again... how can one use S/PDIF to achieve the "DTS/DD" hi-rez audio formats? (you may need to "dumb" it down for some of us. )
I don't know what you mean by "DTS/DD" hi-rez audio. DTS and Dolby Digital are lossy formats and are supported on S/PDIF. That is how people got surround sound on DVD.

Perhaps you are asking about Dolby TrueHD and DTS-MA-HA which are lossless formats used in blu-ray discs. Those are not supported on S/PDIF based on the choice the industry made to hasten adoption of HDMI for audio.

Most players can do on-board decoding of audio in above formats and output said stream on S/PDIF. Alas, AACS copy protection in HDMI forces the sampling to be no more than 48 KHz at 24 bits. While I consider that "high resolution" maybe others do not.

For just about everyone, their source of music enjoyment is not a Blu-ray disc. Yes there are some good titles there but it is a fraction of what I or anyone else would listen to. Those sources are CD and higher resolution formats available in download all of which is able to travel over S/PDIF.

I expect S/PDIF to get replaced with USB as the high resolution interconnect to music servers. With "async" USB "pull" method, we do away with tracking input clock in the DAC. The DAC creates a stable clock and tells the source to just give it data. That is what the architecture of audio systems should have been instead of synchronizing to input on every clock pulse.

BTW, you can rip the audio tracks out of Blu-ray and convert them to FLAC or other lossless formats which you can then put on S/PDIF (and of course USB).

Quote:
Originally Posted by Ratman View Post
I know you are busy with your own forum and well as educating us on AVS, but I would like to get some direction and insight.

Thanks in advance for your answer and continued support on AVS.
Well, there it is.
amirm is offline  
post #800 of 2920 Old 06-20-2014, 12:55 PM
AVS Forum Special Member
 
krabapple's Avatar
 
Join Date: Apr 2003
Location: in a state bordered by Kentucky and Maine
Posts: 5,636
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 385 Post(s)
Liked: 374
Quote:
Originally Posted by urapnes1 View Post
I don't think what I wrote was whining, but rather a fair feeling that someone could take away after reading posts like that. You would never find that kind of repeated derogatory tone in a true professional setting without HR stepping in. I've seen people fired for similar actions. Personally, I'd just start permanently banning people

Not all forums are this way. CNCZONE is pretty well behaved.

As far as how to cut the crap and get back on topic...There is still a chance. I signed on to this thread to try to learn more about jitter. In the past I could swear I heard differences between CD players. I could describe what I was hearing, but never knew what jitter was. Always wanted to know.

Before joining the 'maybe it's jitter' line dance there were and are more likely explanation to explore regarding those differences you 'could swear' you heard.
krabapple is offline  
post #801 of 2920 Old 06-20-2014, 12:57 PM
AVS Forum Special Member
 
krabapple's Avatar
 
Join Date: Apr 2003
Location: in a state bordered by Kentucky and Maine
Posts: 5,636
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 385 Post(s)
Liked: 374
Quote:
Originally Posted by Kees de Visser View Post
Our perception of noise must be very different.


You have to hear 'through' the noise

But is enough dithering going on here for that?

Last edited by krabapple; 06-20-2014 at 01:01 PM.
krabapple is offline  
post #802 of 2920 Old 06-20-2014, 01:14 PM
AVS Forum Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 17,528
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1487 Post(s)
Liked: 1191
Quote:
Originally Posted by amirm View Post
No one can answer that because we do not remotely have enough sampling of HDMI products to know how they perform. You can ask if your HDMI AVR has audible jitter but then you have to present me with its measurements. If we have that, we can try to apply our understanding of psychoacoustics to figure that out.
So... it's a non-issue. Because with your experience and reluctance to provide a definitive answer, what is the constant babble about in this thread?


Quote:
Originally Posted by amirm
Your question is akin to asking me if the fish *you* buy at your grocery store is safe to eat when all you have seen is a sampling of 5 grocery stores elsewhere in the country.
Not relevant.


Quote:
Originally Posted by amirm
I don't know what you mean by "DTS/DD" hi-rez audio. DTS and Dolby Digital are lossy formats and are supported on S/PDIF. That is how people got surround sound on DVD.

Perhaps you are asking about Dolby TrueHD and DTS-MA-HA which are lossless formats used in blu-ray discs. Those are not supported on S/PDIF based on the choice the industry made to hasten adoption of HDMI for audio.
Yes, I think you do. Don't "dumb it down".
What's your alternative to enjoy those audio formats without degradation due to HDMI jitter?

I know those formats are not supported on S/PDIF! That's why I asked the question! How do YOU suggest WE alleviate the inherent "jitter problem" of HDMI and enjoy those formats?

Not even addressing the rest... as usual, "dodge, dip, dive, duck, and dodge"

Last edited by Ratman; 06-20-2014 at 01:21 PM.
Ratman is offline  
post #803 of 2920 Old 06-20-2014, 01:27 PM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by Frank Derks View Post
A higher resoltion file and/ther music selection *might* shift the audibility threshold in this particular test 6 dB lower but that's about it.
This is still much higher than the encountered jitter levels.
What "encountered jitter levels?" How much HDMI jitter instrumentations have you seen?

There is no way you can make such generalizations as "encountered jitter levels." Here is an easy example from my data, this time showing a much larger spectrum:



Both measurements are from the same Anthem processor. The difference is input test signal. One is a straight 12 Khz pure tone. The other, J-test which is also 12 Khz but with the amplitude changing one bit from time to time.

You see very different spectrum of distortions/jitter for the two different input signals. The reason is that the content itself is determining what distortion will be there. We call this "correlated" distortion. Its audibility is much higher than the uncorrelated 30 Hz in Arny's test. The brain will tend to tune out non-variant signals after a while. But distortion that comes and goes can color that specific input that it is going with.

Also note that there are a heck of a lot of distortion products none of which are at "30 Hz." The first significant spike in pink is at 225 Hz and for green at 320 Hz or so for pink. But again, these vary depending on what is being played and the system in question so please don't run with them and put us through more tests that don't extrapolate to the topic at hand.

Let me state the only proper generalization you can make about HDMI performance: in all but the best implementations, HDMI produces more distortion and noise than other digital interfaces. How crappy it becomes is system dependent. And by system I am including the source in there since HDMI electrically couples the two and so you have to worry about the computation.

Now, a lot of this may not be audible and certainly not to all people on all content. So if this is for movie watching and such, by all means go ahead and use HDMI. That is what I do. But if you are building a music server, then use USB to S/PDIF conversion or native USB (if it is implemented properly). You are guaranteed to get less distortion than HDMI. Will it make a difference in your life, maybe maybe not . I am not here to judge but to provide the data. You want to ignore it, please do but don't respond to me saying I am supposed to believe.
amirm is offline  
post #804 of 2920 Old 06-20-2014, 02:12 PM
AVS Forum Special Member
 
Frank Derks's Avatar
 
Join Date: Jul 2002
Location: Region A,B,C
Posts: 2,373
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 459 Post(s)
Liked: 335
Nice try, lumping other distortions together with jitter might seed some doubt with the audiophile clientele but these are what they are **other** distortions.


Nobody denies that hdmi is more prone to jitter and other distortions. To put the attempt to seed FUD into the proper context is once again the level it needs to be to become audible.
Your graph shows it perfectly well. Those are very low level distortions .


The J test result in this graph is the typical misuse to 'highlight' a imaginary 'problem'. Music is not 'changing one bit from time to time'.
Guess how this signal dependent noise floor will look when playing actual content. For example from a properly dithered cd.


Arny's test files showed perfectly the amount of jitter levels in relation to audibility for the isolated case of 30Hz jitter.
For other distortions audibility other test are more appropriate.
Frank Derks is offline  
post #805 of 2920 Old 06-20-2014, 02:25 PM
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,420
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 851 Post(s)
Liked: 1242
Quote:
Originally Posted by amirm View Post
Arny, you created the test. The listeners have no choice but to live with those conditions.
You seriously misjudge me if you want to suggest that I would ignore a good idea about how to improve listener sensitivity in this test.

The fact is that ABX was the result of me listening to listener complaints about why the test methodology we started out with was unnecessarily hard.

The latest set of files are the result of more than 30 years sifting through various ideas about how to actually make these tests easier.

Quote:
Originally Posted by amirm View Post
I am asking about your intentions in creating this specific test. In this thread you have been saying that the lower the frequency of jitter, the more masking makes it potentially inaudible.
Not at all. I've been repeatedly saying that for the jitter frequencies we find in common audio gear, masking doesn't matter and I've cited Zwicker and Fastl (2007) as my authority for that. I've repeated this message again and again and again. I've even posted a link to the whole Zwicker and Fastl book online and downloadable, worth about $80, as substantiation for that.

Quote:
Originally Posted by amirm View Post
In this test, you have picked the 30 Hz frequency.
Actually, 30 Hz jumped out of test results that you have psted here, Amir, both yours and other people's. I've annoted it and posted it and annotated it for my own reference.

Quote:
Originally Posted by amirm View Post
If your logic is correct on masking, which it is , then it reasons that you created a listening test that gave the lowest odds of someone hearing real life jitter. As such it would be improper to draw any conclusions from this one test.
That is a false claim for reasons given above. I've also offered to recreate the files with other jitter frequencies if:

(1) The alternative jitter frequency were provided.

(2) If there were enough people who promised to make a full run of tests and post their results.

Quote:
Originally Posted by amirm View Post
If we really care about audibility of jitter in real products we need to simulate far more frequencies and all at once as the devices output.
Quote:
Originally Posted by amirm View Post
That is insufficient for proper analysis Arny. Signal processing issues may show up in the music segment but not the 1 KHz tone.
Please provide an explanation how that can happen since my tests show that the jitter adding process is signal independent and it is applied to the whole file in one uninterrupted pass.

Quote:
Originally Posted by amirm View Post
Is there a reason you are being secretive and not telling us how the files are created?
I have revealed as much as I think I can without biasing the listeners.

I notice that JA seems happy with his analysis of the test tones in the files. I have provided detailed descriptions of how to analyze the files.

Quote:
Originally Posted by amirm View Post
We ultimately need to know that what we tested is what is purported to be.
It is there in plain sight.

Last edited by arnyk; 06-20-2014 at 02:28 PM.
arnyk is offline  
post #806 of 2920 Old 06-20-2014, 02:27 PM
Senior Member
 
urapnes1's Avatar
 
Join Date: Mar 2007
Posts: 310
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 44 Post(s)
Liked: 39
Quote:
Originally Posted by krabapple View Post
Before joining the 'maybe it's jitter' line dance there were and are more likely explanation to explore regarding those differences you 'could swear' you heard.
Explanations such as: ???????????? -- read below

I "thought" I was able to isolate what I was hearing to a particular CD player. I had 2 cd players hooked up to the same preamp-amp. I had 2 copies of the disc and switched back and forth between the 2 listening for the same passage over and over. I looked at a few magazines (primarily Stereo Review and Stereophile) at the time (early to mid 90's), and most seemed to mention THD and jitter figures (I don't remember what else S/N maybe a few others). Mistakenly, or not, I came to the conclusion that if similar metrics were being used to evaluate the performance of a CD player, then perhaps variations among those metrics could take place and perhaps that what I thought I was hearing could have been, in this case jitter. (I loosely understood what S/N and THD meant, but not jitter). I don't see why this would be a wrong train of thought for someone who was interested in learning more about what they thought they heard. 20 years later the subject came up, so I thought I would listen in. I had no concept of level matching at the time, nor the ability to correct it even if it was off.

As I stated earlier...I came here to learn. Isn't that one of the main purposes of forums like this...to share information so others can learn?
urapnes1 is offline  
post #807 of 2920 Old 06-20-2014, 02:40 PM
AVS Forum Special Member
 
mcnarus's Avatar
 
Join Date: Nov 2007
Posts: 6,156
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 153 Post(s)
Liked: 374
Quote:
Originally Posted by urapnes1 View Post
I "thought" I was able to isolate what I was hearing to a particular CD player. I had 2 cd players hooked up to the same preamp-amp. I had 2 copies of the disc and switched back and forth between the 2 listening for the same passage over and over. I looked at a few magazines (primarily Stereo Review and Stereophile) at the time (early to mid 90's), and most seemed to mention THD and jitter figures (I don't remember what else S/N maybe a few others). Mistakenly, or not, I came to the conclusion that if similar metrics were being used to evaluate the performance of a CD player, then perhaps variations among those metrics could take place and perhaps that what I thought I was hearing could have been, in this case jitter. (I loosely understood what S/N and THD meant, but not jitter). I don't see why this would be a wrong train of thought for someone who was interested in learning more about what they thought they heard. 20 years later the subject came up, so I thought I would listen in. I had no concept of level matching at the time, nor the ability to correct it even if it was off.
It sounds like you tried to do the best you could given the equipment you had and what you knew at the time. That's understandable.

But we are all older and wiser now, and we (and I suspect you) can do a much better job of explaining your experience than we could have back then. For example, you now know (I hope) that both THD and jitter levels in non-defective CD players are typically orders of magnitude below the threshold of audibility.

Also, if you've been hanging around AVS for more than about 10 minutes you know that two CD players can have different output levels, and even very tiny differences in levels can produce what sounds like very obvious differences in sound quality even from otherwise identical sources.

Given all that, you should be able to easily deduce the most likely and least likely explanations for the differences you heard 20 years ago. And if so, you have less to learn than you think you do.
CruelInventions likes this.

If you can't explain how it works, you can't say it doesn't.—The High-End Creed

mcnarus is offline  
post #808 of 2920 Old 06-20-2014, 02:55 PM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by Frank Derks View Post
Nice try, lumping other distortions together with jitter might seed some doubt with the audiophile clientele but these are what they are **other** distortions.
What audiophile clientele??? The only time I have these discussions is on forums where folks want to believe what they want, the actual system design and data be damned.

Other distortions are just that: other distortions. They could be modulation of voltage reference (Vref). If so, their amplitude will not be signal dependent (but whether they exist due to that, can very much be). Vref modulations will create symmetrical sidebands by the way just like clock jitter. So that "nice try" doesn't apply here because you don't know which is which from looking at the graph.

Quote:
Originally Posted by Frank Derks
Nobody denies that hdmi is more prone to jitter and other distortions.
My job here would be one heck of a lot easier if no one denied that. But denying they do. Before performing my own measurements, I would cite the ones from Paul Miller's web site which I also quoted here. During one of those interactions a couple of years ago, Arny says this:

Quote:
Originally Posted by arnyk View Post
As usual Amir you are avoiding the point which is that any competent digital receiver is supposed to resolve any likely data timing issues that arise. If certain obsolete products [in Paul Miller's site] didn't do this, that must be one reason why improved models were produced.

Again Amir you seem to have no up-to-date technical measurements of the HDMI performance of AVRs that are currerntly on the market as new products.
The usual banter on this forum would have said, "why don't you Arny go and test the new products and show us the "data timing issues" have been "resolved." I didn't do that. I thought more data is always good. So I set out to collect a sampling of devices with (and some without) HDMI and ran a suite of tests on them. These are the measurements I have been posting. As you see, they clearly invalidate Arny saying the non-obsolete products have eradicated this problem. If anything it is the other way around. The obsolete "high-end" products wipe the floor with these mass market AVRs.

Going into this test I had no idea what the outcome would be. And I held a neutral position. I gathered the information and published it in a magazine read by the industry and enthusiast alike. There has not been engineer, designer, or even a market person coming forth saying, "hey, you made our product look bad. your measurements are wrong." Or coming forward with another model that doesn't have the problem. Instead I got postings like this from our fellow objectivist member, Ethan on this WBF forum thread:

Quote:
Originally Posted by Ethan Winer
I don't know how many WBF'ers read Widescreen Review magazine, but Amir has been writing a regular column about various audio topics. This month's (March 2013) article about jitter is the best yet. He explains the different types of jitter and their causes, and shows lots of graph to supplement the text. Hopefully Amir will put this article on his own web site after a suitable period, and link to it from here so everyone can see it.

As some of you may know, Amir and I had an extended discussion about jitter a while back, with me arguing that while jitter is real I don't think it's actually an audible problem. Son of a gun, Amir came to the same conclusion in this article, and he even explained why it's not necessarily audible (masking of nearby frequencies). However, I agree that audio designers need to aim higher than "borderline acceptable" for several reasons. For example, audio typically passes through many devices in a row, so if each is just barely transparent the sum of artifacts can be audible.

Anyway, I just wanted to compliment Amir for his good work, and I learned several things myself.

--Ethan
And what is the reaction in this forum across so many threads? Nothing but hostile reception even though the ultimate message is friendly to our "objectivity" cause as Ethan mentions above. No, we can't even give one millimeter to the other side. Let's do everything from insulting to downright misstatement of the technology as to cloud the data. Common sense is hard to find in this forum.

Quote:
Originally Posted by Frank Derks
To put the attempt to seed FUD into the proper context is once again the level it needs to be to become audible.
And here is the best example of above.

Quote:
Originally Posted by Frank Derks
Guess how this signal dependent noise floor will look when playing actual content. For example from a properly dithered cd.
Sorry but I am not into guessing and certainly not into pseudo audio science. But sure, go ahead and test your measurements of that and we can take a look.
amirm is offline  
post #809 of 2920 Old 06-20-2014, 03:03 PM
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,420
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 851 Post(s)
Liked: 1242
Quote:
Originally Posted by amirm View Post
Let's consider Arny's listening tests. As distortions became smaller, I used specific slices of 2-3 seconds to detect differences. Finding and picking such small segments in a track is hard and prone to mistakes. The proper file would be revealing across most if not all of its track, making it much easier to find the artifacts.
When I see comments like this I am very happy because they are symptomatic of a person who is actually reliably hearing differences as opposed to what we hear from high end reviewers and audiophiles who are doing sighted evaluations and are actually just reporting their illusions.

Hopefully we'll keep this on the down low so that people don't start reporting this sort of thing as the outcomes of their sighted evaluations because they can imagine anything they would like.

I call these small segments that seem to be more diagnostic for audible difference Critical Segments. We've been using various features of CD players to isolate and repeat Critical Segments since the feature showed up on the CDP 101 in 1983.

FOOBAR2000 has its Set Start and Set End feature to allow people to exploit this sort of thing. That's the standard implementation of this kind of feature and is IME about as good as it gets.

AFAIK there is no such thing as real world CD tracks that are Critical Segments from end to end. In our dreams!

Perceptual coder developers and investigators collect and share Critical Segments. Some specific ones have been mentioned in magazine articles base on DBTs of perceptual coders.

There are archives of Critical Segments as related to perceptual coder flaws on the web. In general Critical Segments are very specific for the kind of artifact. They may also be somewhat specific to various listeners.

The particular track I used - taken from "Little Green" by Joni Mitchell seemed to be exceptionally diagnostic for jitter in my listening tests. Yes there are parts of it that seem to be better than others, but they were pretty obvious to me and are easy to find near the beginning of the track. I've provided listeners with Critical Segments that are only a few seconds long and got accused of putting the listeners into strait jackets, so I left things a little more open ended.

I'd be glad to find out about other such Critical Segments. Relevant DBT listeners only need apply.

Last edited by arnyk; 06-20-2014 at 03:13 PM.
arnyk is offline  
post #810 of 2920 Old 06-20-2014, 03:35 PM - Thread Starter
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,829
Mentioned: 6 Post(s)
Tagged: 0 Thread(s)
Quoted: 1334 Post(s)
Liked: 741
Quote:
Originally Posted by arnyk View Post
FOOBAR2000 has its Set Start and Set End feature to allow people to exploit this sort of thing. That's the standard implementation of this kind of feature and is IME about as good as it gets.
As good as it gets? I think it is one of the worst implementations. So that everyone can follow us, here is the sample picture of the UI:



First problem is that when you move the slider it does not update the time code. You have to hit the Start or End buttons to see it. This means a bunch of hunting with a slider to try to get it back to the same segment.

The slider cannot be stretched which means that it is very difficult to get it down to tenth of a second which is what you need when you want to isolate say, a single guitar pick.

When you hit the "set start" button nothing happens. It actually remembers that point but it shows no visual notification. It is only when you hit the end button that it shows where the start was. I want to see that marker so that for example I can give myself 1 second after that. Without seeing where the start position is, I wind up with a pain in the neck trial and error.

Then there is lack of looping. It plays from start to end of that segment but won't repeat. You want it to repeat so that you then switch inputs and hear their effects.

The other problem of switching tracks. There is no soft dissolve so you get a click as you switch between the input tracks.

Compounding all of this is that every time you start the tool it forgets it settings so all the pain repeats to adjust the slider.

And then there is the undiscoverable way of saving the results. You have to hit exit and then it gives you a chance to save the results. The file name however is blank. It should instead start with the file names. The ABX box often hides the tracks and when you are in the save dialog box, you cannot move it to see your track names.

All of this make it a lot harder to setup the right test. I gave up and just used the first segment I found. It was just too much work to try to locate others.

So if one wants to have less revealing results, you can say this is a good tool. Otherwise, it leaves a lot to be desired versus ones written by people who do this for a living.
amirm is offline  
Sponsored Links
Advertisement
 
Closed Thread Audio Theory, Setup, and Chat

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off