Blue or red pill?

[fas42]

8 hours ago, Sonicularity said:

What causes confusion for me is the idea that detailed descriptions are provided by the listener about how the sound differs, which would correspond to someone being able to hear differences with the relatively large gap in time between playback of the test samples.  Yet, this same information was lost in the first attempts using ABXXXXXXXXXX.  Immediately I suspect the test procedure is somehow responsible for the differences, creating some tell.   It just seems like a logical step in the process to understand what is happening.  I come across as the bad guy to some when I think along these lines; though, I really just want to learn what is going on. 

 

 

 

Human hearing is remarkably adaptive, and very quick to learn - once you 'know' what the best quality of some piece of music is your brain fills the gaps, and they all sound like the best version. Everyone experiences this type of thing with a poor quality radio - listen to some music you're not familiar with; it sounds terrible - put on a favourite tune, and you'll happily bop away to the sound ... your brain and memory add all the needed oomph ...

 

Unfortunately, human hearing is not like a really, really dumb animal, that can asked to sit, roll over, beg, etc, etc, ad nauseum, and keep playing the game on cue ... scientists get bugged by the fact that people are like, well, people - and are not perfectly predictable, .

[fas42]

14 minutes ago, Audiophile Neuroscience said:

I have no issue with challenging the evidence, looking for uncontrolled variables and so fourth but thus far there appears only IMO rather far fetched speculation of maybe potential possibilities without a shred of evidence to support it.

 

 

Anyone who has experienced how doing "crazy things" affects the sound, like myself, wonders what all the fuss is about - the world is in a fine balance of an enormous number of factors determining everything, and human hearing is just a bit more more senstive to some non-obvious cause and effect chains. Some people prefer the world to be a simple, robotic place; like a well-oiled piece of software - but it 's just a tiny bit more complicated than that ...

[fas42]

3 hours ago, pkane2001 said:

 

It certainly helps a lot to have some test tracks to play with, other than the ones I generate myself!

 

 

 

This is an excellent resource, for test data: https://www.gearslutz.com/board/showpost.php?p=13240356&postcount=1617. Not all the links still work, but there's plenty there to play with.

[fas42]

5 hours ago, Ralf11 said:

I hope you guys will focus on the mechanism for whatever you may have found at whatever likelihood of significance...

 

Pretty obvious, which it makes it remarkable that there is so much thrashing around, in the conversation regarding the possibilities with the hearing. In audio reproduction there is one large circuit producing the sound, usually separated in separate boxes - one could have the audio server, media player, DAC, amplifier all in one box, and then people would say, of course it's obvious that there could be interference effects disturbing the sound audibly. But put everything in separate boxes, and suddenly a great miracle occurs - there is now infinite separation between the various parts, they are all pure, Black Boxes. So, so nice for the objectivists ...

 

I've found the hardest thing is to completely separate areas of functionality, so that in fact they are truly robust in the face of other electrical activity - this is the chasm that has to be crossed to secure competent sound, always. What the precise mechanism is, is vastly less interesting than resolving the linkage - a fully sorted rig should always sound identical, irrespective of what the media player is, and its settings - that's the real goal, if there is to be long term sanity in all this.

 

 

[fas42]

Still doodling with the 9 and 10 waveforms - all manual, I'll leave it to Paul to get a "proper" mechanism to fire up, - upsampled to 18 Meg, to make it easy to adjust alignment, and just looking at above 5kHz audio ... starting to look interesting - there's a spike in the spectrum at about 13k in the diff; and getting some single cycles in the diff waveform peaking very significantly at points where there is nothing particular in the original, at those high frequencies. Confirmed that I'm looking at something real, by slipping one sample on 10, and looking at what that produces when diff'ing.

 

Is this just pure noise, random artifacts from the ADC imperfections? Need to peruse further ...

[fas42]

Sorted ... should have read the next posts, first, .

[fas42]

4 hours ago, manisandher said:

 

I took a listen to the analogue captures I linked to earlier today and they sounded worse than the previous ones from the MOTU ADC. They also seemed to have a slight channel imbalance. I then realised that the MOTU's input was set to 'variable' instead of 'fixed', meaning that the signal from the DAC was passing through additional pots and circuitry in the MOTU.

 

Which is how it works. Every unnecessary part in the circuit path, unless it is 100% pristine, is liable to degrade the sound - this relates to every area of a playback rig; optimising is merely locating the worst offenders, and removing them, replacing them with greatly improved quality units, or redesigning how it works.

[fas42]

Well, the drift has been solved, but the two versions aren't aligned to the beat of the clock. Just looking at one spot, comparing 23 and 24, the samples are out by about 2u secs - resampling will be necessary, to get decent nulling.

[fas42]

32 minutes ago, STC said:

 

 

Tried this and this is so much easier. No fancy system. Just a $20 Sennheiser earphones and PC. But you can only get this right after being familiar to the particular test. Throw me a new track and I am bound to fail. 

 

 

Interesting. For whatever reason, on a cursory listen I found these latest two analogue captures harder to distinguish from each other, compared to the previous lot - I'm pleased that someone can pick up a clearcut variation, makes investigating them more worthwhile.

[fas42]

Okay, before anyone gets too excited - there's a tell at 0.9 sec in, which lasts for 0.025 secs. This could be a glitch in the recording process, for some reason - and is probably enough to 'learn' to recognise. It appears to be the only one, without looking more thoroughly at the difference file.

[fas42]

35 minutes ago, acg said:

 

STC did 50 ABX comparisons in 12 minutes...that's 14 seconds per ABX...it appears as though he could hear that "tell"

 

Located what the "tell" is - before the first piano note sounds, at the beginning, there is a very short snippet of acoustic, about a quarter of a second; from the recording. In one sample, that acoustic snippet is slightly longer than the other, 0.025 secs worth ... now, is that long enough, significant enough for one's mind to register? I certainly can't hear it being special, but other people might pick up on it.

[fas42]

1 minute ago, STC said:

 

You are speculating if you can't tell the difference. Even when we hear identical tracks, we may perceive difference but is it real or imaginary?

 

If one is to be certain that an unconscious trigger is not being fired each time, then the tells need to be excluded. Which is easy to do for Mani - he just needs to adjust the beginnings of the track, so that they match, perfectly.

[fas42]

Just now, STC said:

 

IMO, this approach is wrong. The question is - Is there a difference? If yes - can you hear it? If yes - can you be consistent? It is no point reading into someone mind on how and why they could tell the difference because the difference itself is not constant to matter.

 

But the point is that this difference is purely the result of how Mani organised the track, to upload - absolutely nothing to do with the quality of the replay ... we need to start off by being confident that we are comparing oranges with oranges - and that can be done by ensuring that non-playback related aspects of the samples match.

[fas42]

I can't see or hear this.

 

 

Can you show us where you see it?

 

Mani.

 

In Audacity,

 

 

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

[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 ...

[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.

[fas42]

What happened with the digital captures is that sync with the input was lost, being controlled by an alternative clock. Interestingly, the result was that, at least in one portion of the waveform, that the two digital captures were perfectly in step with each other, then lost the lock, only to re-establish it - with a regular beat. This looked like,

 

 

The flat line links are in fact perfect nulls, the two digital captures match in these intervals.

 

[fas42]

The real silliness in all this is the belief that even a 100% properly done ABX actually 'proves' anything about what's going on with human hearing - there is a great conceit with much of science in the belief that they can always get a handle on how the human organism operates using simple laboratory methods, and there have been some terrible disasters as a result of this type of thinking.

 

I would be quite certain that you could have two people in a room participating in the identical ABX, one who has strong conviction that they can hear the differences; and the other believing it is all nonsense; the former would get excellent results, the latter would achieve almost perfect random scoring - because the need to be right would override anything his senses were telling him ... and absolutely nothing would be gained, yet again.

[fas42]

Even a relatively poor measuring ADC would normally have the ability to register the differences in the output - which is all that is needed. It's not absolute performance that's being measured - but rather, a detectable variation.

 

Part of the Art of Science is to craft a method of observing something, when the measurement tools are "not good enough!" - not much use throwing up one's arms, and saying, "Ye canna do it!!"