Can you review a Synchro-Mesh S/PDIF re-clocker?

[Empirical Audio]

Oh that offset. Fully aware of it. Would you care to explain how it changes the recovered audio data?

If I can identify the offending copies, what does that matter? It proves that this corruption changes the audio output.

 

[March Audio]

If I can identify the offending copies, what does that matter? It proves that this corruption changes the audio output.

Does it? Can you? Really? Please explain what the corruption is to the final audio output.

Data read from the disc has plenty of errors but are corrected by Reed Solomon code.

 

[Empirical Audio]

Not sure. However, jitter in this context is similar to a raised noise floor.

http://www.sereneaudio.com/blog/what-does-jitter-sound-like (look at random not periodic jitter)

Take note that what they call 2 microsecond of random jitter, because of the different sample rate and maybe dithering doesn’t follow the calculations I did earlier, as I did my calculations on non-dithered and 44.1kHz. Also, their graphs shows ~75dB of dynamic range, but if you take the dynamic range test I linked above, it sounds much worse than that, so something it wrong; but again, talking magnitudes more than what is typical in real-world applications.

I've read random jitter tests before. I don't believe they are relevant. What's needed is a correlated jitter to mimic real music jitter.

 

[MZKM]

What's needed is a correlated jitter to mimic real music jitter.

I wouldn’t think random jitter would be any more forgiving than real-world jitter. You can’t give real-world jitter as test signals either (you would have to be able to have real-world systems with poor jitter and capture it from the analog out of the DAC), but then the jitter of the user system comes into play.

Again, this is why we have the J-Test. There is no jitter in the test itself, but it provides a worst case scenario with the LSB and all that, so it should be impossible for real-world jitter to be worse than it.

 

[Killingbeans]

By listening. This is not a hobby where we just measure. It's about listening to reproduced music. You are just trying to bait me.

After this testing is done, I want to prove to everyone reading here that listening tests are just as important as measurements. Both are required until we have a measurement method and equipment that correlates 100% to what we hear. That correlation simply does not exist.

It just amazes me that you seem to find the combination of your so called 'highly resolving' setup and a persons auditory system to be infallible.

Is there absolute zero risk of this measurement method being so elusive, solely because the perceived improvement of SQ is a figment of your imagination?

If there's no correlation between two methods of collecting data then how can you automatically assume that one of them is without fault? Shouldn't both be scrutinized further?

Sorry, if I seem to be baiting, but I could just as well accuse you of trolling

 

[svart-hvitt]

https://en.m.wikipedia.org/wiki/Herd_behavior

 

[amirm]

I don't draw the line. I continuously improve the system by rooting out the things that affect SQ.

That reminds of a story I have told but the newcomers probably have not heard it.

When I was managing the audio codec team at Microsoft, I trained my hearing to detect very small artifacts that my team could not. So in spare time I would spend time testing new releases of our audio encoder prior to release. I was home sick for a week and the manager of the codec team asked me about a new encoder they were developing. I tested it and gave them feedback but the back and forth was slow. So I asked them to give me control of the parameters so that I could tune them and listen back and forth. So they did and gave a configuration file with a dozen or so parameters. I started to tune them and realized that even changing the fractions in those numbers made a distinct change in fidelity. So I sat there, hours and hours and tuned the numbers to a few decimal places.

Once I was done, with lots of proud, I hand over the configuration file to my codec manager in email. He responds with surprise saying that the software doesn't care about the fractions. I said that is completely wrong and that it absolutely did. That I could change a fraction and AB encoded files before and after and there was clear evidence of one being more distorted/harsher than the other.

Next he emails me two sets of files and asks me which is better. I listen and one set was atrocious. It was clearly worse sounding. Between the cold and him challenging me this way, I expressed my frustration to him that they could not hear such clear differences. He then responds with with a short answer: the two sets of files he had sent me were identical!

I could not believe it. So I do a binary comparison of the files and they were identical to the last bit! I had heard clear difference that were absolutely not there. I do a comparison then and all of a sudden, all the differences I thought were there, were gone! Man was I embarrassed.

The story doesn't end there. I wondered how it could be that I hard heard such a difference. So I did a thought experiment of imagining the files were different. Immediately I could hear the difference again! This occurred even with full knowledge that the files were different.

So be careful of thinking you are continuously improving things. Pause and run a controlled, blind test. It is so easy to chase your tale thinking what you physically improved, must have had an audible difference. Your ear+brain are one of the most unreliable tools here when detecting the results of the work you are doing. That is just the way they work. It is called being human. Our brain is programmed to make us optimistic. It is dying to find reasons to celebrate. If it were the other way around, we would all die of pessimism and negative thoughts!

 

[amirm]

Have you ever read a Stereophile review and compared JA's measurements to the listeners assessment? JA uses the same methods and AP system that Amir does, right?

He doesn't actually. Not that it matters but he has the last generation 2700 Audio Precision analyzer. I have the newer generation. The 2700 is an excellent analyzer though and is only bested slightly by mine:

Mine is the APx555 in black and his is 2700 in blue. Graph is from Audio Precision.

 

[Newk Yuler]

That reminds of a story I have told but the newcomers probably have not heard it.

Not necessarily a unique testimony but still awesome. I need to print this post, frame it and put it on my office wall.

DAMN I'm loving this thread.

 

[tpaxadpom]

He doesn't actually. Not that it matters but he has the last generation 2700 Audio Precision analyzer. I have the newer generation. The 2700 is an excellent analyzer though and is only bested slightly by mine:

Mine is the APx555 in black and his is 2700 in blue. Graph is from Audio Precision.

As of this year 2700 series analyzers are considered legacy products. I would not be surprised if JA will switch to APx555 in the near future.

 

[amirm]

As of this year 2700 series analyzers are considered legacy products. I would not be surprised if JA will switch to APx555 in the near future.

It will be a good test of their new management to see if they will spend this kind of money on testing or not.....

 

[FrantzM]

It will be a good test of their new management to see if they will spend this kind of money on testing or not.....

Considering the price of the gears they routinely "review" they should. Perhaps they can find some kind of "Industry accommodation" with AP.

 

[KSTR]

In theory. However, in practice it does not matter how good a job you do on the USB interface, the things I mention will still have an effect. There is never zero jitter, but other things come into play with USB including common-mode noise on the USB signals and software sensitivities.

Just add a SOtM USB regenerator to the USB cable and see what a difference that makes.

Steve N.

I did, using the Intona USB isolator and different USB hubs, as proposed by Archimago in his blog. No difference with the RME Adi-2 Pro that I could detect in a blind test. Also I could not detect the difference of the small digital processing bit error that I found in the RME's firmware vs. when corrected (you may have seen my thread here), and this error I would naively assume to have a much larger impact than any jitter.

In sighted tests I tend to prefer whatever is (in my thinking) technically more correct without doubt. Simply a result for my engineering-driven personal biases that I do not (and cannot have) any control of in sighted tests. You cannot just say "OK, now I will listen to this with no expectations, completely unbiased". It does not work, at least for my lowly tin-eared self anyway.

 

[PierreV]

It is really hard for me to imagine how bad a system must have sounded at the start one can make 7 or 8 "clearly audible improvements" (router, ethernet cables, isolators, reclocker, audiophile switch, modded amplifier, speaker cables, etc...). That's even more surprising when one considers that such systems' main initial components are usually already highly rated in subjective or even objective reviews and have often gone through multiple iterations that improved on previous generations...

There is such a statistical bias towards improvement in reviews, manufacturer, and users' claims that a benevolent audiophile god seems to be guiding our journey. Even well-known gods don't seem to have that success rate.

 

[kiat]

after 22 pages of comments, has any equipment been measured yet?

 

[Frank Dernie]

Have you ever read a Stereophile review and compared JA's measurements to the listeners assessment? JA uses the same methods and AP system that Amir does, right?

I have.
My opinion is that it shows that lots of people like the sound of distortion, though I still think the placebo effect is by far the most plausible scientific explanation, all others are harder for me to believe, anyway.

 

[Frank Dernie]

It will be a good test of their new management to see if they will spend this kind of money on testing or not.....

Doesn’t he have a loan unit from AP?

 

[Empirical Audio]

Does it? Can you? Really? Please explain what the corruption is to the final audio output.

Data read from the disc has plenty of errors but are corrected by Reed Solomon code.

For these files, we are not talking about reading from a disk or any correction.

 

[Empirical Audio]

I wouldn’t think random jitter would be any more forgiving than real-world jitter. You can’t give real-world jitter as test signals either (you would have to be able to have real-world systems with poor jitter and capture it from the analog out of the DAC), but then the jitter of the user system comes into play.

I would think that correlated jitter is more obvious than random jitter because the brain is focused on specific music occurrences and the brain already knows what these sound like in real life. If there is any deviation from what the brain knows, this will be easy to discern. Random jitter is like noise events that occur randomly. There is no way for the brain to determine whether this is in the music or not.

We need to use real music for jitter tests, not J-Test.

 

[Empirical Audio]

It just amazes me that you seem to find the combination of your so called 'highly resolving' setup and a persons auditory system to be infallible.

I never claimed that. I only know that my own auditory system is pretty well trained after 38 years. Other people are certainly more susceptible to what I refer to as "going down the garden path".

Steve N.