Blue or red pill?

[manisandher]

7 hours ago, fas42 said:

I can't hear it either - but since we were mentioning tells, it was a point that some might consider significant.

 

Ah OK, this was the ~0.03s of difference in silence between the two that I mentioned earlier. Got it.

 

Mani.

[manisandher]

8 hours ago, adamdea said:

I think...

 

I'm not interested.

 

8 hours ago, adamdea said:

... time to move on.

 

Please do.

 

Mani.

[Sonicularity]

2 hours ago, manisandher said:

 

The listener has a name. And how the f**k do you know what I was listening for?

 

Mani.

 

In the first and second ABXXXXX.. tests, you would have heard A, B, and then X in the first iterations.  This was the same as your individual ABX tests where you were 9/10 in your responses.  I was simply pointing out that the first response was missed in the first two ABXXXXX... tests, yet from the information that was provided, the first response should have been identical to the other ABX tests..  

[esldude]

Did I get a error in download or do the digital captures not match in the most recent batch of files.  I see they are a different length, but don't appear to null out even prior to that endpoint when matched up.  

 

You get periods of full nulling with periodic glitches in the result.  Here is an Audacity view of the files once aligned at the beginning with one inverted and the pair mixed together.  The digital captures are off.  Assuming the analog captures were experiencing the same thing no wonder they are heard in an ABX comparison.  

 

The glitches digitally are occurring about every 20,000 samples or roughly 1/8th of a second.  Each glitch lasts for about 14,000 samples with the 6000 or so samples between fully nulling out. 

 

[manisandher]

35 minutes ago, esldude said:

Did I get a error in download or do the digital captures not match in the most recent batch of files.

 

OK, perhaps my idea of slaving the Tascam's wordclock to the MOTU's wasn't very good after all:

 

On 29/04/2018 at 7:47 PM, manisandher said:

I have no idea what happens when a clock drifts during a digital capture, i.e. no conversion taking place. I suspect nothing much, provided the drift is within a specified range. But I felt that slaving the digital capturing device's clock to that of the analogue capturing device couldn't do any harm, and might perhaps prove useful for alignment.

 

Perhaps I should have slaved the MOTU to the Tascam? But I felt that the wordclock should sit as close to the ADC chip as possible.

 

In any event, the analogue captures should be totally unaffected.

 

Mani.

[mansr]

15 minutes ago, manisandher said:

OK, perhaps my idea of slaving the Tascam's wordclock to the MOTU's wasn't very good after all:

 

On 4/29/2018 at 7:47 PM, manisandher said:

I have no idea what happens when a clock drifts during a digital capture, i.e. no conversion taking place. I suspect nothing much, provided the drift is within a specified range. But I felt that slaving the digital capturing device's clock to that of the analogue capturing device couldn't do any harm, and might perhaps prove useful for alignment.

Yeah, you can't do that. The digital receiver must use the same clock as the sender. With S/PDIF this is achieved by letting the receiver recover the clock from the signal. If you force it to use a different clock, bad things happen.

[lmitche]

1 hour ago, manisandher said:

 

OK, perhaps my idea of slaving the Tascam's wordclock to the MOTU's wasn't very good after all:

 

 

Perhaps I should have slaved the MOTU to the Tascam? But I felt that the wordclock should sit as close to the ADC chip as possible.

 

In any event, the analogue captures should be totally unaffected.

 

Mani.

How about we record the DAC output to analog tape and later cut vinyl for remote particpants? We could even put track A on one side and track B on the other! 

 

I heard somewhere this has been done before. It should be feasible.

[Ralf11]

that LP should have a Red side and a Blue side 

[STC]

2 hours ago, esldude said:

Did I get a error in download or do the digital captures not match in the most recent batch of files.  I see they are a different length, but don't appear to null out even prior to that endpoint when matched up.  

 

You get periods of full nulling with periodic glitches in the result.  Here is an Audacity view of the files once aligned at the beginning with one inverted and the pair mixed together.  The digital captures are off.  Assuming the analog captures were experiencing the same thing no wonder they are heard in an ABX comparison.  

 

The glitches digitally are occurring about every 20,000 samples or roughly 1/8th of a second.  Each glitch lasts for about 14,000 samples with the 6000 or so samples between fully nulling out. 

 

 

I didn’t compare the digital but for some reason when I inverted A and B of the analogue there was practical no difference in Sound. It looks they are very different from each othe although the waveforms looked almost identical. Audacity broken?

 

 

[fas42]

2 hours ago, manisandher said:

In any event, the analogue captures should be totally unaffected.

 

Mani.

 

Just to confirm, there appears to be no obvious glitches in the analogue captures. That Diff file shown in the earlier post was normalised, and played: the differing length of the initial acoustic produces a distinctive zzzt, but from then on there is just the original track, buried in noise. 23 and 24 slowly drift in alignment, so the music slowly gets louder until the end.

 

I haven't worked on getting a finer alignment yet, to deepen the null - this hopefully will show something useful ...

[Audiophile Neuroscience]

 

Adam you have a peculiar knack for missing the point and misrepresentation

 

 

12 hours ago, adamdea said:

 

I think it's a shame that their appears to be no scope for reasoned discussion.

 

I can't reason you out of a position you didn't use reason to arrive at in the first place.

 

 

12 hours ago, adamdea said:

 

Equally, ever since the original 9/10 was reported, you have wanted to see this as proof of one proposition only, on the assumption that it must mean what you think it does, there apparently being (for you) only one explanation.

 

No, that is your unfounded assumption. I have repeatedly stated I am open to getting at the truth wherever it may fall. I have repeated stated this is not about proof. Whether you just misconstrue things or mischievously misrepresent the matter I cannot say.

 

 

 

8 hours ago, adamdea said:

The only real point of contention is that for some reason you consider it less plausible that a tell should be affected by the change in test methodology than that the alleged software effect.

 

The reason is as previously stated.

 

 

 

8 hours ago, adamdea said:

 it was perfectly apparent that all that could possibly be implied (aside from learning) was that the change in test methodology affected the ability to detect the tell.

 

NO that is not all that could "*possibly* be implied" nor is it "perfectly apparent". If you have evidence to the contrary of course lets hear it.

 

8 hours ago, adamdea said:

The reason why the tell might work on the ABx  but not ABxxx

 

So now its a "might". Make up your mind, is it "perfectly apparent", "all that could be possibly implied" or just a "might".

 

8 hours ago, adamdea said:

the ABx the listener was called upon to compare B with X which the listener can do, whereas in the ABxxxx, he couldn't really as A and B were too old. That still leaves the question of what exactly he is detecting in the ABx comparison which he could not detect in the ABXXXX.  I made this point right at the beginning. There is no reason I can see why a tell could not be affected by audio memory just as much as whatever the software is supposed to do.

 

It is my contention (NO, NOT proof) that the tell was likely not dependent on memory in the same way as the sound quality comparison in relation to software change. One fairly unlikely (yes my opinion) proposed possible tell was hearing clicks, or timings of clicks, through two closed doors and a corridor.However if it could be heard, recalling the sound of a simple keystroke or obvious timing gap would, I submit, place very little strain on auditory memory.If such a thing occurred I believe most of us could remember and identify the tell. Complex musical passages on the other hand are something quite different.

 

8 hours ago, adamdea said:

In turn I find it implausible that any audible software  effect could be incapable of being recorded. And I happily stand by that.

 

You are entitled to that opinion. I am a little more skeptical but absolutely open to the possibility.

 

[esldude]

21 minutes ago, STC said:

 

I didn’t compare the digital but for some reason when I inverted A and B of the analogue there was practical no difference in Sound. It looks they are very different from each othe although the waveforms looked almost identical. Audacity broken?

 

 

No there is an intersample timing shift.  The remaining sound had the characteristic 6 dB per octave upward tilt. The general level I think was down about 15 db combined this way.

[fas42]

23 minutes ago, STC said:

 

I didn’t compare the digital but for some reason when I inverted A and B of the analogue there was practical no difference in Sound. It looks they are very different from each othe although the waveforms looked almost identical. Audacity broken?

 

 

 

Simple, the uploaded analogue captures are opposite in phase to the digital capture, as Mani recorded them. If one is sensitive to phase, then it should be obvious in the hearing.

[STC]

I think I got so much more to learn. Now, I wonder if I made other mistakes with other files. Thanks guys.

[pkane2001]

Here are some preliminary results using my (still in early development) DeltaWave audio viewer:

 

First, the original two files (13A and 14B, both analog, 1 second removed from both ends of both files):

 

 

Then, the same waveforms after drift correction, gain matching, and phase alignment:

 

Now, spectrum (FFT) analysis of the original two files:

 

Spectrum after drift correction, gain matching, and phase alignment:

 

And finally, the spectrum of the two, fully matched files, subtracted from each other:

 

(The X-Axis units for waveform data are in samples, for FFTs it's the frequency in Hz.)

 

[STC]

5 minutes ago, pkane2001 said:

Here are some preliminary results using my (still in early development) DeltaWave audio viewer:

 

First, the original two files (13A and 14B, both analog, 1 second removed from both ends of both files):

 

 

Then, the same waveforms after drift correction, gain matching, and phase alignment:

 

Now, spectrum (FFT) analysis of the original two files:

 

Spectrum after drift correction, gain matching, and phase alignment:

 

And finally, the spectrum of the two, fully matched files, subtracted from each other:

 

(The X-Axis units for waveform data are in samples, for FFTs it's the frequency in Hz.)

 

 

And that means......?  Identical?

[pkane2001]

13 minutes ago, STC said:

 

And that means......?  Identical?

 

Looks so, except for some slight differences above 17KHz, which I doubt are audible, but maybe?

 

I'll start comparing other captures, just to see if there are differences that might show up once in a while.

[pkane2001]

27 minutes ago, pkane2001 said:

 

Looks so, except for some slight differences above 17KHz, which I doubt are audible, but maybe?

 

Here's the portion above 17Khz showing some divergence between A and B captures:

 

 

[manisandher]

2 hours ago, pkane2001 said:

I'll start comparing other captures, just to see if there are differences that might show up once in a while.

 

That looks like a really neat piece of software you've created there.

 

Could you compare the digital files 21 and 22 at some point please? It'd be interesting to see if you get the same result as @esldude.

 

Mani.

[esldude]

3 hours ago, pkane2001 said:

Here are some preliminary results using my (still in early development) DeltaWave audio viewer:

 

First, the original two files (13A and 14B, both analog, 1 second removed from both ends of both files):

 

 

Then, the same waveforms after drift correction, gain matching, and phase alignment:

 

Now, spectrum (FFT) analysis of the original two files:

 

Spectrum after drift correction, gain matching, and phase alignment:

 

And finally, the spectrum of the two, fully matched files, subtracted from each other:

 

(The X-Axis units for waveform data are in samples, for FFTs it's the frequency in Hz.)

 

Looks very interesting.  So do you need some beta testers? ?

[pkane2001]

10 minutes ago, esldude said:

Looks very interesting.  So do you need some beta testers? ?

 

Soon!

 

[STC]

9 hours ago, pkane2001 said:

 

Looks so, except for some slight differences above 17KHz, which I doubt are audible, but maybe?

 

I'll start comparing other captures, just to see if there are differences that might show up once in a while.

 

Interesting that you mentioned about difference above 17kHz.  I am still trying to hear the difference the original analogue A and B. After many attempts I sense slight tonal difference and decided to increase the treble to see if I could hear the difference better.  The best I could do was 6 correct in a row. I usually get the first 4 correct but I find it impossible to concentrate long enough to endure the 10 attempts.

 

Do you think Mani's extra 10 ft cable and amplifier would have magnified the difference?

[pkane2001]

9 minutes ago, STC said:

 

Interesting that you mentioned about difference above 17kHz.  I am still trying to hear the difference the original analogue A and B. After many attempts I sense slight tonal difference and decided to increase the treble to see if I could hear the difference better.  The best I could do was 6 correct in a row. I usually get the first 4 correct but I find it impossible to concentrate long enough to endure the 10 attempts.

 

Do you think Mani's extra 10 ft cable and amplifier would have magnified the difference?

 

It's interesting that I do see the differences above 17KHz, and they seem to become more substantial (more than a few dB) as the frequency increases. While I doubt that these differences in frequency response are audible by themselves, some of Mani's analog equipment can possibly be reacting to the higher frequencies in a way that reflects back into the audible range. Can't prove it, since we don't have captures from the output of the amplifier/speaker system. And if we did, these wouldn't be very easy to compare in any case.

 

[pkane2001]

8 hours ago, manisandher said:

 

That looks like a really neat piece of software you've created there.

 

Could you compare the digital files 21 and 22 at some point please? It'd be interesting to see if you get the same result as @esldude.

 

Mani.

 

I'll do that, Mani. Looks like I need to add one more graph, though, one showing the result of subtracting the two waveforms after alignment to see what Dennis saw. The FFT plots and original waveforms look very closely matched, with just a slight mismatch in level of 0.2dB and a tiny offset in phase of about 1 sample or 6μsec.

 

What was different about how 21 and 22 were captured that resulted in differences in level? Both of these were digital captures, right?

[STC]

21 minutes ago, pkane2001 said:

 

It's interesting that I do see the differences above 17KHz, and they seem to become more substantial (more than a few dB) as the frequency increases. While I doubt that these differences in frequency response are audible by themselves, some of Mani's analog equipment can possibly be reacting to the higher frequencies in a way that reflects back into the audible range. Can't prove it, since we don't have captures from the output of the amplifier/speaker system. And if we did, these wouldn't be very easy to compare in any case.

 

 

 

As I previously mentioned, some difference can be more audible with speakers system due to their distance. High frequency can easily be damped with air and the difference of over the speakers can be audible as being slightly mellower. I find it easier to distinguish MP3 over CD with my main system then with headphones. 

 

The HF can be further attenuated by the cable and speakers. IMO, the best way to show whether difference exist that can be audible in a given system is to record them with a microphone at the listening spot.