Blue or red pill?

[PeterSt]

2 minutes ago, STC said:

still unable to find a similar chart to Man's .

 

Figure 9 (and 8) in the link I gave are about that (though the spreading base is not emphasized there). But you will see that the graphs are "even" at both sides. What sticks out are the "side lobes" (at various places).

 

I can tell you that looking at such static pictures (be it those in the PDF or the one from Mans) is not really educative. Only seeing it happen in real time gives a good idea what it is about. Unless you know the theories already. But hey, chicken-egg stuff eh. Maybe you can Google YouTubes for 

jitter fft

 

 

[STC]

13 minutes ago, PeterSt said:

I can tell you that looking at such static pictures (be it those in the PDF or the one from Mans) is not really educative. Only seeing it happen in real time gives a good idea what it is about. Unless you know the theories already. But hey, chicken-egg stuff eh. Maybe you can Google YouTubes for 

jitter fft

 

 

Thanks Peter. 

[STC]

Quote

 

What kind of insane hardware depends on CPU timings and doesn't have a hardware buffer that always plays at the same speed? Jitter in the intermediate steps doesn't matter when you're moving digital data from one place to another, as long as the buffers are larger than the amount of jitter.

DAC is hard. Moving a couple megabits of data five feet over a cable with multi-millisecond buffers is not hard.

I don't think it's even possible to use USB for 100%-throughput unbuffered data. I call bullshit on the eye pattern affecting playback.

 

 

My earlier posts were referring to the difference that was mentioned here about how recorder and DAC treat the data stream. Hope this is relevant to the discussion.

 

[PeterSt]

37 minutes ago, STC said:

My earlier posts were referring to the difference that was mentioned here about how recorder and DAC treat the data stream. Hope this is relevant to the discussion.

 

It is not about that (looking at that small portion of a seemingly larger thread). Whether the eye (pattern) is larger or smaller or whatever shape, as long as all is error free there are no errors and thus that is not the cause of anything we talk about here (and that is in the context of all being/staying bit identical which *thus* proves that the communication (say eye pattern) is OK - no errors).

 

It is also not about jitter as such in the solely digital domain because jitter in there won't do a thing, when observed from start (PC) to end (DAC's receiver).

 

It is about how the timing differences anywhere (thus also in the pure digital domain) imply different current (spike) usage in analogue parts, like the D/A process itself, right after that (current to voltage conversion when applicable, gain stage (with possibly various speedy parts like opamps)).

 

[STC]

7 minutes ago, PeterSt said:

It is also not about jitter as such in the solely digital domain because jitter in there won't do a thing, when observed from start (PC) to end (DAC's receiver).

 

It is about how the timing differences anywhere (thus also in the pure digital domain) imply different current (spike) usage in analogue parts, like the D/A process itself......

 

I firmly believe in the maxim that "Little bit of knowledge can be dangerous thing " and I am guilty of that. I always taught jitter is a form of timing difference errors.

 

I myself doubt the slight jitter could be audible but you offered a plausible explanation that the jitter (timing differences) may alter other part of the signal(?) .  That much I gather from this 76 page thread.  

[PeterSt]

5 minutes ago, STC said:

I myself doubt the slight jitter could be audible

 

As far as my own experience goes, and this includes the knowledge of applying certain oscillators with (Phase Noise) specifications, upgrade the (commercial) DAC to better specs oscillators, measuring jitter to the extent still possible, without and with galvanic isolation ....

 

of course each improvement to less jitter is audible. Always and vastly.

So what the general person can not see and must trust from hoaxing vendors is that his abc DAC now sounds better because of low jitter and what not. They usually don't even have the equipment to measure (but outsource one set of measurement for the nice commercial aiding) ...  but over here we make the lot ourselves and also have the equipment (say like Stereophile et al uses for supporting reviews (some times contradicting)). 

 

So trust me, if you see the HUGE difference Mans showed (never mind even he himself thinks it is small), it will be easily audible. HOW that is audible I can't predict. But bass is always first (and this is logical).

[STC]

7 minutes ago, PeterSt said:

 

As far as my own experience goes, and this includes the knowledge of applying certain oscillators with (Phase Noise) specifications, upgrade the (commercial) DAC to better specs oscillators, measuring jitter to the extent still possible, without and with galvanic isolation ....

 

of course each improvement to less jitter is audible. Always and vastly.

So what the general person can not see and must trust from hoaxing vendors is that his abc DAC now sounds better because of low jitter and what not. They usually don't even have the equipment to measure (but outsource one set of measurement for the nice commercial aiding) ...  but over here we make the lot ourselves and also have the equipment (say like Stereophile et al uses for supporting reviews (some times contradicting)). 

 

So trust me, if you see the HUGE difference Mans showed (never mind even he himself thinks it is small), it will be easily audible. HOW that is audible I can't predict. But bass is always first (and this is logical).

 

With respect, i disagree that the difference was huge. Out of 30 trials, Mani only got 17 over 30 correct.  In the last 10 trials, where Mani scored 9/10. a different method was used. In the earlier 20 trials, 24/176.4 version of 'Tchaikovsky Hopak from Mazeppa' on Reference Recordings' 'Exotic Dances' file was used. In the last 10 trials, 'Persephone' from Patricia Barber's 'Mythologies' album 16/44.1 file was upsampled to 176.4.

 

Did Tascam capture the upsampled file or the original 16/44.1? 

[PeterSt]

11 minutes ago, STC said:

but you offered a plausible explanation that the jitter (timing differences) may alter other part of the signal(?)

 

Yes, but with a very fast job. The Lush thread (also coincidentally about USB) goes into that much further. But it is too large to find back. Summarized this is about how a USB cable sounds the most different than any other USB cable, and it is exactly about eye openings etc. and how they surely can influence sound when larger or smaller - all error free. We should not repeat that here, but I noticed the subject in that link you gave - it is really the same subject. Very briefly, think like this :

 

You have a cable. This cable, including its transmitter and receiver parts, will incur for a degree of reflection; An up-going cycle, meant to be received at the other end as a binary 1, isn't absorbed fully, reflects back to the transmitter and from there it is reflected back again (to the DAC). Both arrivals at the receiver (mind you, this sits in your DAC) are "converted" by the receiver (say it goes through some chip) and that conversion itself requires (minute) current (notice that "convert" is virtual but it makes understanding more easy). The original signal is received as the binary 1 and the reflected signal is not received as a binary 1 because too weak to pass the threshold of being (interpreted as) a 1. Still it requires current because it still goes through that chip. This in itself is already sufficient to understand how cables in the digital domain incur for audible differences IF you understand (or like to see) that this very small current spike at the DAC's end, influences the analogue parts in it (via the general ground plane etc.). OK ?

That is the Lush application (with secrecy).

 

Now we are going to influence the timing of those reflections, or better : their intensity which is related to timing as well, already because there will be more than one back and forth reflection.

Now you might say "hey, yes, hmm".

No.

It is about oscillations. So where the "false" voltages may pile up and do that in oscillating format. Thus instead of (say) 480 million times per second such a single reflection occurring (which may be audible, but I don't think so), now once per milli second all reflections coincidentally pile up and THAT is audible (and even controllable).

 

These theories go way further because that false voltage spike (current draw) which influences the analogue parts, will also influence the mentioned chip itself (via the very same means). And *that* implies the real oscillation.

So notice that it is oscillation which implies patterns. And patterns we humans are good at.

[PeterSt]

2 minutes ago, STC said:

With respect, i disagree that the difference was huge.

 

With even more respect, I am listening to these differences 3-4 hours a day.

And I really don't need to be present at such a test. As long as XXHighEnd was "conducting" it. And it was ...

 

That Mani couldn't do it in absolute sense ... I already told that although hard to imagine, I probably would have failed that too. But not when I just listen to music (no A-B whatsoever) in unconscious fashion and with further the same test (is that an SFS of 0.1 or is it 200 ?). But I also said : it hugely depends on the music, and I would NOT let pick the music by someone else. Not even the genre. I'd need to pick that myself.

 

So trust me, such differences are huge. And also trust me that I am not able to A-B (I just physically can't, which incorporates psychologically  finding it a waste of time). Just play different tracks (in full) of different albums and let me cook a dinner meanwhile.

[Audiophile Neuroscience]

20 minutes ago, STC said:

 

Out of 30 trials, Mani only got 17 over 30 correct.  In the last 10 trials, where Mani scored 9/10. a different method was used. In the earlier 20 trials, 24/176.4 version of 'Tchaikovsky Hopak from Mazeppa' on Reference Recordings' 'Exotic Dances' file was used. In the last 10 trials, 'Persephone' from Patricia Barber's 'Mythologies' album 16/44.1 file was upsampled to 176.4.

 

 

 

It is not valid to combine the two different tests. The first test based on AB XXXXXXXX failed but IMO this reflects the poor sensitivity of the test and propensity to produce false negatives and false positives. The second test of ABX ABX etc succeeded to a p-value of 0.01 and is highly significant. The second test was equally blinded and in every other way just as rigorous scientifically at determinig a potential difference if one existed.You cannot change the test methodology mid stream and combine the results. Different test procedures must be looked at independently and on their individual merrits. Its like saying we did 20 xrays and found no brain tumour; we did one MRI scan and found a tumour, therefore the tests show 20:1 against having a brain tumour - it doesnt work that way.

[sandyk]

50 minutes ago, PeterSt said:

But not when I just listen to music (no A-B whatsoever) in unconscious fashion

 

 Too much booze ?

 

BTW,  and you call your USB cable Lush.

Guilty by association ?

 

Urban Dictionary: Lush

https://www.urbandictionary.com/define.php?term=Lush

1. Cached

Either sex can be a lush. It's someone who drinks a lot and enjoys drinking a lot---not necessarily an alcoholic, but someone whose drinking is noticeable. It can be used teasingly or as an insult, but even if someone teases you (or whoever), there's probably some truth in it, and it probably means you're overdoing the ...

[dean70]

Does the format (DSD or PCM) show any bearing on susceptibility to jitter/signal reflections or other unwanted noise from the PC? Is DSD more/less immune than PCM or the same in a given setup?

[mansr]

10 hours ago, pkane2001 said:

Mans, can you also post a wider plot of the same data? At least a KHz on each side, and may be the full 20-20KHz range?

Sure.

 

 

Here's another plot showing a minor difference:

[mansr]

3 hours ago, STC said:

With respect, i disagree that the difference was huge. Out of 30 trials, Mani only got 17 over 30 correct.  In the last 10 trials, where Mani scored 9/10. a different method was used. In the earlier 20 trials, 24/176.4 version of 'Tchaikovsky Hopak from Mazeppa' on Reference Recordings' 'Exotic Dances' file was used. In the last 10 trials, 'Persephone' from Patricia Barber's 'Mythologies' album 16/44.1 file was upsampled to 176.4.

The first run of 10 used the Tchaikovsky track. The second run, as well as the third with the altered sequence, were with the Patricia Barber track.

 

3 hours ago, STC said:

Did Tascam capture the upsampled file or the original 16/44.1? 

The Tascam captured what was sent to the DAC.

[STC]

7 minutes ago, mansr said:

The Tascam captured what was sent to the DAC.

 

I probably didn’t phrase the question correctly.

 

The original file was 16/44.1. If you upsampled them to 176 then Tascam should have captured the updampled version of 176. Was that the case?  

[mansr]

15 minutes ago, STC said:

I probably didn’t phrase the question correctly.

 

The original file was 16/44.1. If you upsampled them to 176 then Tascam should have captured the updampled version of 176. Was that the case?  

As I said, the Tascam captured the data sent to the DAC. If that data was the result of upsampling, then that's what was captured.

[STC]

4 minutes ago, mansr said:

As I said, the Tascam captured the data sent to the DAC. If that data was the result of upsampling, then that's what was captured.

 

Ok. I thought the files were bit identical to the original 16/44.1. It is not bit identical to the original files. Correct?

[pkane2001]

57 minutes ago, mansr said:

Sure.

 

 

Here's another plot showing a minor difference:

Thank you, Mans! 

 

Interesting. There is quite a bit of noise with some noise spikes rising nearly 40dB above what appears to be the noise floor. In the zoomed in portion all the spikes appear to be at 100Hz interval, is that right? Seems a bit strange, even for a 50Hz mains frequency.

 

Do you have a calibration scan of a 10KHz sine wave using your ADC? Something we could compare these scans with without Mani's PC and DAC in the circuit?

 

[adamdea]

On 4/10/2018 at 1:19 PM, Audiophile Neuroscience said:

 

 

What specific "tells" needed to be further controlled for and what are the supposed "conclusions" you claim I am leaping to?

 

1) The same ones I have already mentioned eg clicky changes, possible single blind problems whatever. The usual things people do with potentially surprising experimental results

 

2) That the 9/10 involved detection specifically and only of the difference between the audio output of a dac fed with the  bit identical output of two pieces of software. Obviously. In one or two places you make moderate comments about food for thought, but most of the time you are singing Ding Dong the Wicked Witch is Dead (out of key too). It's boring to keep repeating oneself but the caveat is obvious to anyone- what exactly was being detected? This is basic stuff; it shouldn't even need saying.

[mansr]

3 minutes ago, pkane2001 said:

Thank you, Mans! 

 

Interesting. There is quite a bit of noise with noise spikes appear to rise nearly 40dB above what appears to be the noise floor. In the zoomed in portion all the spikes appear to be at 100Hz interval, is that right? Seems a bit strange, even for a 50Hz mains frequency.

Yes, there's a spike every 100 Hz. No, that's not normal.

 

3 minutes ago, pkane2001 said:

Do you have a calibration scan of a 10KHz sine wave using your ADC? Something we could compare these scans with without Mani's PC and DAC?

We used Mani's ADC. What would you use as a reference?

[adamdea]

14 hours ago, mansr said:

And somehow the actual test ended up being something quite different.

Well quite.

[pkane2001]

16 minutes ago, mansr said:

Yes, there's a spike every 100 Hz. No, that's not normal.

 

We used Mani's ADC. What would you use as a reference?

 

Ah, I didn't realize it was Mani's ADC. A sine wave generator, if you have one. If not, any well-known, quality DAC (preferably with measurements) with a digital sine wave generator to feed the ADC. Worst case, any other DAC, again preferably with measurements.

 

[adamdea]

5 hours ago, STC said:

 

With respect, i disagree that the difference was huge. Out of 30 trials, Mani only got 17 over 30 correct.  In the last 10 trials, where Mani scored 9/10. a different method was used. In the earlier 20 trials, 24/176.4 version of 'Tchaikovsky Hopak from Mazeppa' on Reference Recordings' 'Exotic Dances' file was used. In the last 10 trials, 'Persephone' from Patricia Barber's 'Mythologies' album 16/44.1 file was upsampled to 176.4.

 

This part troubles me. Aside from my concern about switching tells, I'm not sure whether this should really be treated as a multiple comparison test.  It would be interesting to run again with a pre-determined specification of what the test was and what would constitute success. 

[jabbr]

15 hours ago, PeterSt said:

 

http://www.analog.com/media/en/technical-documentation/application-notes/AN-756.pdf

I never read that one, but the Analog Devices application notes are usually quite easy to read. It can't harm to browse through it.

?

 

Trying to wade through this long an interesting thread. I am most interested in the conclusions of the participants @manisandher, @mansr As well as you, the author or software and designer/manuf of hardware. 

 

Is there a summary thread? 

 

Is there data which supports lower close-in phase error? (narrower peaks)

[manisandher]

20 hours ago, mansr said:

And somehow the actual test ended up being something quite different.

 

19 hours ago, mansr said:

The test was of some mystery setting in a playback program, not of streaming vs local storage.

 

6 hours ago, adamdea said:

Well quite.

 

Are you serious?

 

I first started mentioning how we'd conduct the A/B/X back in Feb:

 

On 14/02/2018 at 7:19 AM, manisandher said:

... However, I was intending to do the A/B/X purely in Attended mode, which would make the A/B/X easier to perform. I've checked and I can clearly hear the difference between SFS=20 and SFS=1, which should suffice for the A/B/X.

 

I elaborated on this further at the beginning of March:

 

On 02/03/2018 at 1:06 PM, manisandher said:

That's a specific case of the more general hypothesis: a file played* back bit-identically can sound different.

 

Specific cases then include:

- different storage media

- different digital cables (spdif, USB, etc)

- different software player configurations (buffers, etc)

 

I'm considering which of these would be best for the ABX, and am leaning towards different software player configs. Once you're satisfied that there really is an audible difference between a file played back bit-identically (whether you hear it too, or whether I manage to prove it in the ABX), then we can certainly explore the other specific cases, if you have the time and inclination.

 

The day before the visit, I outlined to Mans in a PM exactly what we'd be doing:

 

"This is the procedure I'd like to use tomorrow:

 

1. take a quick listen together

- I'd like to demonstrate a few things to you and get your initial thoughts

 

2. conduct the A/B/X

- I've chosen the track and the bit-identical changes we'll use in the playback software

- you'll be sitting in my office, controlling playback from there, and I'll be sitting in the listening room

- we'll have the Tascam set to auto-record, sitting in the basement next to the audio PC and DAC, capturing the digital output of the audio PC in real time

..."

 

If anyone had any concerns with how bit-identical playback was going to be achieved, they had ample time to voice them.

 

Mani.