Quote

And here are the files for you to do DBT:
I'd put two files at my server with 30kHz and 15kHz FR limits.
[a href="index.php?act=findpost&pid=236598"][{POST_SNAPBACK}][/a]

What is the relevance of ABXing a 15kHz low pass? Is it just to demonstrate that a previous experiment (which showed a 15kHZ low pass was inaudible) was wrong? I don't think we need to prove that again - there are enough threads already! Anyway, the experiment wasn't wrong - it showed that those listeners with those equipment listening to that particular sound couldn't detect a 15kHZ low pass filter. It's almost irrelevant to high quality music listening - drawing the conclusion that 15kHZ was enough for music was a false conclusion, but it's the applied logic that's faulty, not the scientific results.

As for the CD/SACD/DVD argument, none of them use a lowpass of 15kHz, so again, what are you trying to prove here?

Cheers,
David.
[a href="index.php?act=findpost&pid=236638"][{POST_SNAPBACK}][/a]

If you believe that, then consider it already broken. Their have been numerous 'tests' that established audibility WITHOUT closely analysing the source signal, ampllfiers, transducer distortions.....

Well, you seemed to not like the idea of non music signals. So with that NHK lab test using music signals, you can't use that as an objection.

No, we don't know. That's the point. We don't know what differences(errors) the filter may have caused in the software(careful analysis of test waveforms needs to be performed first, in the software before/after this filtering), the possibility of hardware distortion(transducers will very probably do this if the ultrasonic amplitude is anywhere near the ampllitude of the audible band content, see earlier referenced Griesenger presentation), possibility of error by person prepping the file(s), etc.

I have no intention of ABXing the files you presented(becuase of the above reasons), but I did download and the first thing I noticed is that you already made one error. ONe file is mono, while the other is stereo. Errors are easy to make. An error free test is very difficult to pull off. That's why I don't give much respect to such loosely controlled tests such as your trying to perform. No where near the level of controls are being used to near scientific validity.

What is the relevance of ABXing a 15kHz low pass? Is it just to demonstrate that a previous experiment (which showed a 15kHZ low pass was inaudible) was wrong? I don't think we need to prove that again - there are enough threads already! Anyway, the experiment wasn't wrong - it showed that those listeners with those equipment listening to that particular sound couldn't detect a 15kHZ low pass filter. It's almost irrelevant to high quality music listening - drawing the conclusion that 15kHZ was enough for music was a false conclusion, but it's the applied logic that's faulty, not the scientific results.
As for the CD/SACD/DVD argument, none of them use a lowpass of 15kHz, so again, what are you trying to prove here?

I'm not impressed by the grammar of the claim , and even less by the lack of scientific listening data in that white paper.  No description of listening test methods, controls, qiuantiation and statistiical analysis of results...just a dubious claim that 'warmth, speed (!) and smoothness were the key words described by members of the listening panel'.  I already told you what my criteria for convincing comparison reports are..this doesn't even come close. 
So, so far, neither of your links supports your claim, AFAIK.

Depends on what the modification is, and what the evidence is for audible difference.  They deserve respect as far as their claims are supported by good reasoning and empirical evidence, no more, no less. It's quite possible -- I've seen it many times -- that smart people can believe wrong things as well as right things ...hey, it's happened to me once or twice too.   Chris Johnson, I am familiar with, and he *generally* takes care to support his claims....but I've also seen him get into some huge wars with objectivists over on RAHE overs some of his claims.   Have any of htse people reported results of listening tests for particular mods?

Page doesn't open.. does it report properly done listening tests?

Measurably differences do not necessarily translate into audible ones.  That's a foundational concept in thisforum... which is why , here, unlike 99% of forums about audio, claims of audible difference have to be
backed up by proper empiricial tests.

I can understand and accept you scientific approach - everything which is not confirmed scientifically and correctly documentated, doesn't exist   All people who got their devices modified, just fool themself with the difference in audibility, because they have no scientific confirmation about it, right?

None of them is "correctly documented", therefore they don't exist. 

Explain to me the condition of "properly done listening tests"

Quote

I can understand and accept you scientific approach - everything which is not confirmed scientifically and correctly documentated, doesn't exist   All people who got their devices modified, just fool themself with the difference in audibility, because they have no scientific confirmation about it, right?

Quote

None of them is "correctly documented", therefore they don't exist. 

It is not a matter of scientific approach, it is a matter of forum rules

It's in the FAQ. Direct link : http://www.hydrogenaudio.org/forums/index....showtopic=11442
[a href="index.php?act=findpost&pid=236757"][{POST_SNAPBACK}][/a]

Within the HA forum rules, you're allowed to quote measurements, describe objective improvements, and speculate.
What you can't do is say that "this sounds better" without ABXing the difference.
The placeabo effect is just too strong to do things any other way. As has been proven thousands of times.

btw, the author of the paper you've linked to, Malcolm Hawksford, was my PhD tutor. If I learnt anything from him (and I learnt a lot - I hope!), it's that nothing is perfect, and the basic circuit theory that most people are taught is full of simplifications that hide all the faults!

In general, he thought that the weakest link in the audio system was often the recording itself.

We both heard and commented on some very bad and very good "high resolution" (i.e. DVD-A or SACD) demonstrations. The best was this one from David Chesky.

100% agree with that. But if we can't say "it sounds better" it doesn't mean that such a difference can't be existent, right? It can be present, just has not been confirmed by DBT yet.

Yes and no ... IMO. We still have 48kHz DSP processing in majority of AV processors, receivers and DVD players, cheap old DACs, noisy clocks, cheap slow op amps with high level of distortions, etc.  ... "Often" - is a right word

OK, then forget about it  We don't have the correctly documented scientific researches concerning difference in audibility of op amps, AFAIK, therefore nothing else will satisfy you. I can understand and accept you scientific approach - everything which is not confirmed scientifically and correctly documentated, doesn't exist 

All people who got their devices modified, just fool themself with the difference in audibility, because they have no scientific confirmation about it, right?

None of them is 'correctly document', therefore they don't exist

Quote

Page doesn't open.. does it report properly done listening tests?

Unfortunately, no - then you don't need it, right?

Quote

Measurably differences do not necessarily translate into audible ones.  That's a foundational concept in thisforum... which is why , here, unlike 99% of forums about audio, claims of audible difference have to be
backed up by proper empiricial tests.

Explain to me the condition of "properly done listening tests" ... especially taking in consideration that for my wife and my son there is no difference between a boombox and HT system in sounding ...

Get 100 of such people and you will have 0 difference in audibility in "properly documented listening tests", then let's all use boomboxes?

My son can't hear the difference (both are satisfactory for him, he just doesn't care) between the regular CD quality and CD, created from 64kb/s WMA file ... Audible difference is a subjective matter ...

you can get some statistical information, but it is not relative to the abilities of personality to hear the difference ...

Quote

Quote

I can understand and accept you scientific approach - everything which is not confirmed scientifically and correctly documentated, doesn't exist   All people who got their devices modified, just fool themself with the difference in audibility, because they have no scientific confirmation about it, right?

Quote

None of them is "correctly documented", therefore they don't exist. 

It is not a matter of scientific approach, it is a matter of forum rules

Thank you, Sir. I had been invited to this forum by krabapple to discuss the article. It looks that I have to stay in AVS forum, their rules are more suitable for me, as I am not a scientist. I simply like to make some assumptions and discuss them with knowledgeable people, therefore this forum with it's rules is not for me.
The rules of it doesn't even allow to discuss the modification of devices or any DIY experience without the DBT ...

Quote

It's in the FAQ. Direct link : http://www.hydrogenaudio.org/forums/index....showtopic=11442
[a href="index.php?act=findpost&pid=236757"][{POST_SNAPBACK}][/a]

Thank you.
[a href="index.php?act=findpost&pid=236762"][{POST_SNAPBACK}][/a]

I invited you here because I knew that you'd get high-quality feedback on how to proceed in your research.  That high-quality is largely due to the enforcement of stringent 'objective' standards for claims...it keeps the signal high and the noise low.   You'd be hard pressed to find a set of people more knowledgeable on how to verify the existence of an audible difference, this side of a psychoacoustics lab. There's also more than a few apparently competent engineering types here.
Now, if we could only get Nika Aldrich (sp) to join the thread ...  that guy ran the best damn 48 vs 96kHz discussion I've ever seen on the Internet.

Quote

100% agree with that. But if we can't say "it sounds better" it doesn't mean that such a difference can't be existent, right? It can be present, just has not been confirmed by DBT yet.

The probability that an actual difference is not found in a DBT is determined by its type II error, which is dependent not only on the parameters of the test but also on what percentage of the population can distinguish a difference.

Just plugging in some numbers into ff123's spreadsheet, I haven't really done this math myself: For type 1 error of 5%, type 2 error of 10%, and 50% of the hifi "population" capable of telling a difference in a blind test, one needs ~31 listeners. This, coincidentally (or not?), is roughly equal to the number of NHK respondents (33). If one then assumes that only a fraction of the population (say, 1%) can tell a difference in a blind test, the number of required respondents for type 2 error=10% jumps to 85,634.

It means that if NHK labs measured hearing abilities, or conscious mind response, and Oohashi measured the brain activity (i.e. unconscious mind reaction), they could get the different results.
[a href="index.php?act=findpost&pid=244970"][{POST_SNAPBACK}][/a]

IMO p=1 may be true for night-and-day kind of differences, but certainly not for subtle audio differences. Your power calculations are only correct under this two-kinds-of-listeners assumption! Do you really think that the highly controversial ultrasonic content is night-and-day kind of difference?

You are also making the assumption that each individual performed exactly one trial. This is not what happened in the NHK experiment, which was a partially repeated measures design, and somewhat flawed in how results were analyzed. This discussion is getting very OT, so I’ll start another thread under listening tests.
[a href="index.php?act=findpost&pid=253676"][{POST_SNAPBACK}][/a]

Quote

IMO p=1 may be true for night-and-day kind of differences, but certainly not for subtle audio differences. Your power calculations are only correct under this two-kinds-of-listeners assumption! Do you really think that the highly controversial ultrasonic content is night-and-day kind of difference?

Well, the context of my reply was that "night and day" is exactly what audiophytes are claiming, and it would be a start to show that this is not the case, that it must necessarily be at most a very subtle effect. IOW, in order to disprove the argument "SACD/DVDA is obviously better than CD for all listeners", all that is necessary to disprove is H3:p>0.9 or so for a group of trained listeners, and especially under a two-kinds-of-listeners model (which everybody seemed to be operating under anyways). That alone would deflate a lot of people's arguments for hires....

I'd rather not get into the details of a proper statistical test, because as far as I know, one would need to test for roughly 0.5<p<0.55 which is obviously hard. It seems a lot more economical to, instead of conduct a statistical test for audibility, just find somebody for which the difference is audible. I don't think that such a person has been conclusively found yet.
[a href="index.php?act=findpost&pid=253917"][{POST_SNAPBACK}][/a]