[pinnahertz]

Caveats are that some effects could sink in over a period of time when you're not under pressure, but even if that were true, how would you prove it?

Headaches and the feeling of being under pressure indicates you're trying too hard. There is no time limit for any comparison except of that you place on it yourself. If you feel there might be more difference by listening longer, then by all means do so!

Also perhaps your brain 'fills-in' after a while. Usually I seem to think the second one was better, something I've noticed before in myself with small differences. Also the DAC I used tends to smooth things over compared with the one in my CD player even at 44.1.

Remember, in an ABX test we are not trying to determine preference. If you are, you've just added an additional amount of pressure to yourself. You're trying to discern a difference, and pair X to either A or B, not try to figure out which is better.

See, this is exactly what I was trying to warn about: a rapid, forced, high pressure test, then trying to extract something the test is not designed for, worsening the situation.

I've done some searching on blind tests, and yes, you're right. It seems most to all people can't discriminate on blind comparison and neither can I in the situation I have described. Sorry if I've wasted your time.

Blind comparisons in general? No, not true at all, they do show up audible differences that "most or all" people can discriminate. Perhaps not in a sampling rate comparison, but that's actually expected. There are plenty of things that can show very high percentage results, though.

There are errors of a few uS in one zero crossing point I looked at between the various files when examined in Audacity. In fact re-sampling entirely removes one small cycle through the zero point I looked at. I haven't looked for patterns in time error, but we're looking at a file created presumably with a wideband LPF.

Leaves me wanting to repeat the analogue vs ADA test and wondering if it was for real!

If the test was presented by someone selling a product, it wasn't real.

 

[SimonPac]

Blind comparisons in general? No, not true at all, they do show up audible differences that "most or all" people can discriminate. Perhaps not in a sampling rate comparison, but that's actually expected. There are plenty of things that can show very high percentage results, though.

No, certainly not blind comparisons in general. Just this test. Having said that, when I'm not concentrating and just listening to music I've mistaken which headphones I've had on!

 

[pinnahertz]

No, certainly not blind comparisons in general. Just this test. Having said that, when I'm not concentrating and just listening to music I've mistaken which headphones I've had on!

Yeah, I know. Kind of disturbing, isn't it? We've all been fooled...Like listening all day to .mp3 files and enjoying them.

 

[SimonPac]

If the test was presented by someone selling a product, it wasn't real.

That's certainly a possibility. They could have down sampled below 44-16 and up sampled for example.

I accept the 44.1-16 vs Hi Res non-significance certainly on A/B based on my gear/material and other tests. Jury is out for me on good analogue vs same analogue to 44/96/192-16/20/24 and back.

 

[castleofargh]

That's certainly a possibility. They could have down sampled below 44-16 and up sampled for example.

I accept the 44.1-16 vs Hi Res non-significance certainly on A/B based on my gear/material and other tests. Jury is out for me on good analogue vs same analogue to 44/96/192-16/20/24 and back.

with any analog recording media that would be audibly different, it would probably be faster to just measure a few stuff and find out what's wrong on the analog side. I know you're not talking about vinyl and K7 tapes, but most of the time the fidelity issue is still on the analog side. but of course nowadays most measurement rigs aren't fully analog anymore ^_^. I guess you can interpret that as an unfair test.

but anyway, your test isn't the be all end all of proof. it's wrong to fully rely on gut feelings, but it's not better to take one test on one rig under one set of conditions, and start drawing conclusions about everything. IMO ABX is a very good test to let us put our feet on the ground. you mention mistaking the headphone you have on. that is indeed the level of erroneous judgements we are all capable of, and have all been victim of. it just goes to show how much nonsense can pass unnoticed as long as the brain is following its own idea unchallenged.

as for the test being tiring, it's fairly typical of first timers, you just put too much strain on yourself. you can try the test with casual level of focus instead of "I must succeed or my family will not eat tonight!" kind of pressure we really like to put on ourselves for no reason. do 4 or 5 runs then stop for a day and resume the next if you feel like it until you get to 20 or 30(whatever number you had decided you would do before starting). and you also decide on your level of commitment. when enjoying music you'd never focus the way you do in the abx, wouldn't it be more realistic to relax a little and forget about that mad desire to get it right? there is no right, if you pass you pass, if you don't you don't, at best it tell you about yourself under those conditions. there is no winning or losing, and while obvious, sometimes it can take time to convince our own brain.
when we enjoy music, even in so called critical listening while sighted, our brain never for a second considers the task to be first priority. music is a fun thingy, not one of those survival stuff. if you were able to focus on music for an entire song the way you did with your abx, or the way you would if you were the next guy to be killed in a scary movie scenario, listening to music wouldn't be associated with the idea of relaxing. ^_^ so with that in mind, think about ABX and what you're really trying to achieve. and while it's never going to become fun, ABX has no reason to be stressful. you're the one in command of everything.

 

[gregorio]

[1] Caveats are that some effects could sink in over a period of time when you're not under pressure, but even if that were true, how would you prove it? Also perhaps your brain 'fills-in' after a while.
[2] I've done some searching on blind tests, and yes, you're right. It seems most to all people can't discriminate on blind comparison and neither can I in the situation I have described. Sorry if I've wasted your time.
[3] There are errors of a few uS in one zero crossing point I looked at between the various files when examined in Audacity. In fact re-sampling entirely removes one small cycle through the zero point I looked at. I haven't looked for patterns in time error, but we're looking at a file created presumably with a wideband LPF.

1. Your brain does definitely "fill-in" after a while but that "while" can be anything from a tiny fraction of a second up to decades! There has been considerable research which demonstrates a two stage memory and that what is stored in short term (echoic) memory, which lasts only for a few (up to 10) seconds, is far more accurate than that stored in long term memory. In fact, I'm unaware of ANY reliable evidence to the contrary.
2. It hasn't wasted our time! Typically, audiophiles are so entrenched in their beliefs that they cannot concede anything which might contradict those beliefs. Provide evidence or proof and they will simply disregard it; if double blind tests provide that evidence then the only logical explanation is that DBTs must be faulty, in order to avoid questioning their unquestionable belief. The result is typically just a downward spiral of disputing overwhelming reliable evidence with effectively nothing more than unquestionable belief, a stand-off with no possibility of resolution. Compared to that more typical result, what's happened so far with your case is not a waste of our time.
3. Audacity certainly has been known to create errors where there shouldn't be any. It's not used professionally for this and other reasons. However, what you've seen may not be an error, it maybe the expected behaviour. Remember that in Audacity and most other DAWs, what you're seeing is a stair-step waveform like graphical representation of the data, NOT what the waveform will actually look like after reconstruction, this fact can actually cause a great deal of confusion.

G

 

[SimonPac]

I've reached the limit of my commitment to AB testing for the moment. I will probably do some informal stuff from time to time. Interesting experience and I needed to do it to adjust my prejudices, so thanks. I'll perhaps try the HD Tracks demo again.

The Audacity waveform display linear interpolates and does so from the stored PCM samples I expect. You don't see staircases in the display whatever the rate. The event I saw was in the file as supplied by HDTracks or whoever it came from, Audacity at 44.1 removed it because it slipped through the re-sampling rate. There actually seems to be a low level 40KHz oscillation or ringing on the 192 waveform! Not good, I think! Whether you would see it in post filter analogue at 192 would depend on the digital reconstruction filter bandwidth and the final analogue LPF. It is removed by the 44.1 process but I don't see why it should be there in the first place.

 

[gregorio]

There actually seems to be a low level 40KHz oscillation or ringing on the 192 waveform! Not good, I think! Whether you would see it in post filter analogue at 192 would depend on the digital reconstruction filter bandwidth and the final analogue LPF. It is removed by the 44.1 process but I don't see why it should be there in the first place.

This sort of thing is not uncommon when using 192kHz. There're quite a few things which can cause interference and it can easily slip through the net during recording and production because of course we (the engineers and artists) can't hear it, we only become aware if we happen to load a section of the audio it into spectogram analysis, which we don't tend to do as standard procedure unless we hear a problem! That sort of problem is largely avoided at 44.1kHz of course. Also, as far as I'm aware, it's still impossible to get the -120dB attenuation from the reconstruction filter which is attainable at 44.1. AFAIK, about -80dB is the max at 192kHz which means some alias imaging slipping in there but at -80dB it should be inaudible. Nevertheless, it's one of the reasons why 192kHz is technically inferior to much lower rates.

G

 

[PETEBULL]

I'd say mechanically overdampened headphones are prone to more distortion at high volume and non-linearity.

 

[71 dB]

The Audacity waveform display linear interpolates and does so from the stored PCM samples I expect. You don't see staircases in the display whatever the rate.

Linear interpolation is of course wrong (but fast to calculate for the screen), as these should be curves build up of sinc waves. The sound is different from your DAC.

 

[71 dB]

I'd say mechanically overdampened headphones are prone to more distortion at high volume and non-linearity.

Why?

 

[SimonPac]

I don't know, people talking about headphone damping? Actually, interesting comment.

Also, as far as I'm aware, it's still impossible to get the -120dB attenuation from the reconstruction filter which is attainable at 44.1. AFAIK, about -80dB is the max at 192kHz which means some alias imaging slipping in there but at -80dB it should be inaudible. Nevertheless, it's one of the reasons why 192kHz is technically inferior to much lower rates.

Stopband attenuation depends entirely on the filter FIR implementation. Not a function of frequency. The only reason high rate would be inferior would be practical limitations on implementation. Hugely faster DACs are used in other fields....

Linear interpolation is of course wrong (but fast to calculate for the screen), as these should be curves build up of sinc waves. The sound is different from your DAC.

True. And the sinc function energy width in time depends on the filter cutoff frequency. Higher sample rate is unambiguously better here because it allows a higher b/w Nyquist filter. Time domain sinc is the way to go for closest approach to ideal brickwall (infinite stopband attenuation) in the frequency/phase domain, if not necessarily optimum in the time domain. There are other ways to do it as we have discussed. Lots of them. Many early Japanese players used complex analogue filters. At least they were causal in playback time. Made me wonder if anyone tries to pre-process re-sampled data to anticipate the new filter cutoff likely to be encountered during replay. Unless I missed something, ideally they should...

 

[PETEBULL]

Why?

Drivers receive more physical resistance and self-interference. All of my highly dampened phones don't like to be loud.

 

[SimonPac]

The ringing on the 192 file I looked at could be 40KHz leaking in electromagnetically into the analogue and not being filtered out by a 22KHz bwf or it could possibly even be a badly designed 40KHz or so nyquist filter or even an oscillating analogue stage. Not an exhaustive list.

 

[71 dB]

[1] Drivers receive more physical resistance and self-interference.
[2] All of my highly dampened phones don't like to be loud.

[1] Resistance is a linear process so it doesn't increase non-linear distortion. It is also likely to reduce linear distortion in a system by making phase-response more resistive, in other words making group delay more constant. Please explain "self-interference" and why damping creates it.
[2] Highly damped phones are less sensitive than less damped phones meaning you need more power for the same sound pressure level. Your highly damped phones "not liking to be loud" is propably about your amp being unable the feed enough power to them. If so, your problem is no damping, but lack of amp power.