Subjective/objective testing /AB / AB/X, thoughts.

Page 3 - Seeking answers? Join the What HiFi community: the world's leading independent guide to buying and owning hi-fi and home entertainment products.

CnoEvil

New member
Aug 21, 2009
556
14
0
Visit site
fr0g said:
.I think many who do the "beating" (me occaisionally, when in the wrong mood) do it becuase they don't like to see people being misled or ripped off. I agree that sometimes the tone (actually most times) is all wrong, and more likely to cause someone to "dig in". Gentle persuasion imo is always best, and if someone "really" isn't interested then it's time to let go.

I would rather that people got both sides of the argument (preferably put respectfully) and were then left to make up their own mind, using whatever method they like.

Despite what many of the "Foo Fighters" would claim, there is (IMO) a genuine debate to be had. People like me, get slated for even suggesting that people try for themselves. I have done (and been to) extensive demos, where I heard clear differences, but since I can't prove it, I never claim that this is anything other than my opinion.

I welcome the chalange of putiing my side of the debate......I'm confident enough of my position that I know there is a good chance that anyone prepared to try it, will hear the difference, if they choose the right product.

Almost every regular on here knows my position, so I won't bore you all with it any more (on this thread anyway).

FWIW. I usually have no problem with your forthright and mostly respectful approach.
 

fr0g

New member
Jan 7, 2008
445
0
0
Visit site
I think one thing that has been shown to me here, is that if a discussion point that often causes conflict, is taken out and discussed on its own merits is less likely to lead to the squabbling we often see. We all have opinions, and nobody is "right".

I do think in some circumstances, blind testing, esp AB/X, is incredibly useful, if only to prove something to yourself, for or against, but in the end I also agree that ultimately, all we can do is sit down and listen, and ask ourselves "Is this nice or what??"

Cheers folks.
 

davedotco

New member
Apr 24, 2013
20
1
0
Visit site
Busy day at the cricket, so have only just got in and read the thread. I have a number of thoughts on the subject, not entirely consistent but hear goes.

Taking part in almost any reasonably well conducted blind test is extremely instructive for any enthusiast, I am not necessarily suggesting you buy your equipment on the outcome of such tests, just that they can give you a very swift grounding of just how small differences can be.

I feel that an experienced listener can, over a period, make judgements on equipment that goes beyond what can be learned from blind testing, that said I think you need to have a lot of experience to make those calls.

One of the few people who have this experience is the dealer, someone good can save you a lot of time and hassle, probably money too. Sadly there are few such that opperate at the budget end of the price spectrum.

Don't try to hard to discount or try to overcome bias, sometimes you can use it to your advantage, buying a better component because you prefer it for some reason other than pure sound quality, can increase your satisfaction by making you more 'comfortable' with it.

For me it really is not an either/or situation, simply being aware of some of the realities of the way hi-fi equipment works and is demonstrated can help you make better informed decisions, that for me is the biggest advantage that objective knowledge can bring to subjective listening.
 

fr0g

New member
Jan 7, 2008
445
0
0
Visit site
John Duncan said:
I think this is an interesting question, which deserves some discussion. I don't believe fr0g is trying to start a fight (though no doubt one will start anyway).It is my opinion that this is a subjectivist forum. In other words, I think people who come here and stay here take the view that the best way of judging a piece of audio equipment is by listening to it. I share that view, but am not immune to the idea of more scientific testing (of which listening is, I believe, not an example, bit more of that later).I think 'formal testing' by means of, for example, ABX is about the best way of deciding whether one thing is different from another (notice that I do not use the word 'better'). However, the ease by which one can achieve this can vary enormously, depending on what it is that one is trying to ABX. For example, I think the tools exist whereby one can satisfy oneself quite easily of the superiority or otherwise of, say higher resolution audio over standard CD or 320k or lower resolution compressed files. I have never tried the Foobar ABX plugin, primarily because I've managed to satisfy myself by more prosaic means (listening) that they are pretty much functionally equivalent (for my purposes). I'd urge people to try it, simply because it's an easy way of taking a position on how one feels about what is 'good enough' from a source point of view.However, I think when it comes to individual pieces of audio equipment, it's much harder to do. All you need to do is look at Alan Shaw's Harbeth Challenge to see the angst that it has engendered in so many quarters. Personally, I think that challenge was never going to be answered (let alone met or beaten) because it was not about hifi equipment at all really, but any respondent's ability to come up with an instantaneous switching mechanism so that an ABX test could be carried out against equipment that doesn't really lend itself to the practice. That, combined with the fact that several 'caveats' about equipment's behaviour were put in place meant that it veered too much towards the old 'all amps sound the same provided you stick a graphic EQ in line with them' question.And there is the nub of the problem - I think it's actually difficult, if not impossible, to ABX hifi equipment (without starting to add other things into the mix like comparators and suchlike, which immediately bring claims of 'signal degradation' which invalidate the tests). It takes too long to swap out pieces of equipment to get round the 'aural memory' issue which pure objectivists claim is the reason behind all sorts of reported phenomena, such as burn-in. And TBH, I'd argue that sticking a sheet over a speaker so you don't know what it is probably counterproductive.So in absence of a practical, repeatable test method, I'm inclined to go with what's left, which is listening. Obviously we can try to remove variables which can affect the result, such as levels, but other than that when I'm looking for a new piece of kit I'm inclined to consider very few questions:- Does it sound better TO ME (since nobody else is buying it for me)?- Is it more functional?- Does it look better?In other words, if I want to spend my money, then I shall, but I'm not so stupid as to declare that the only thing which influences me in that decision is sound quality (since for the last 20 years I'd argue that improvements have been marginal at best, though the cost to implement has decreased fairly inversely to Moore's Law). I will also argue that until somebody comes up with a foolproof, easy method of proving to me (in a nice environment with good coffee, as opposed to a lab or a shed) that X is 1,000 times better than Y I will, as they are forced to do (and are - somewhat curiously - happy to do), have to make do with listening and my own, foolish, subjective opinion.

I think this is an interesting question, which deserves some discussion. I don't believe fr0g is trying to start a fight (though no doubt one will start anyway).

It is my opinion that this is a subjectivist forum. In other words, I think people who come here and stay here take the view that the best way of judging a piece of audio equipment is by listening to it. I share that view, but am not immune to the idea of more scientific testing (of which listening is, I believe, not an example, bit more of that later).

I think 'formal testing' by means of, for example, ABX is about the best way of deciding whether one thing is different from another (notice that I do not use the word 'better'). However, the ease by which one can achieve this can vary enormously, depending on what it is that one is trying to ABX. For example, I think the tools exist whereby one can satisfy oneself quite easily of the superiority or otherwise of, say higher resolution audio over standard CD or 320k or lower resolution compressed files. I have never tried the Foobar ABX plugin, primarily because I've managed to satisfy myself by more prosaic means (listening) that they are pretty much functionally equivalent (for my purposes). I'd urge people to try it, simply because it's an easy way of taking a position on how one feels about what is 'good enough' from a source point of view.

However, I think when it comes to individual pieces of audio equipment, it's much harder to do. All you need to do is look at Alan Shaw's Harbeth Challenge to see the angst that it has engendered in so many quarters. Personally, I think that challenge was never going to be answered (let alone met or beaten) because it was not about hifi equipment at all really, but any respondent's ability to come up with an instantaneous switching mechanism so that an ABX test could be carried out against equipment that doesn't really lend itself to the practice. That, combined with the fact that several 'caveats' about equipment's behaviour were put in place meant that it veered too much towards the old 'all amps sound the same provided you stick a graphic EQ in line with them' question.

And there is the nub of the problem - I think it's actually difficult, if not impossible, to ABX hifi equipment (without starting to add other things into the mix like comparators and suchlike, which immediately bring claims of 'signal degradation' which invalidate the tests). It takes too long to swap out pieces of equipment to get round the 'aural memory' issue which pure objectivists claim is the reason behind all sorts of reported phenomena, such as burn-in. And TBH, I'd argue that sticking a sheet over a speaker so you don't know what it is probably counterproductive.

So in absence of a practical, repeatable test method, I'm inclined to go with what's left, which is listening. Obviously we can try to remove variables which can affect the result, such as levels, but other than that when I'm looking for a new piece of kit I'm inclined to consider very few questions:

- Does it sound better TO ME (since nobody else is buying it for me)?

- Is it more functional?

- Does it look better?

In other words, if I want to spend my money, then I shall, but I'm not so stupid as to declare that the only thing which influences me in that decision is sound quality (since for the last 20 years I'd argue that improvements have been marginal at best, though the cost to implement has decreased fairly inversely to Moore's Law). I will also argue that until somebody comes up with a foolproof, easy method of proving to me (in a nice environment with good coffee, as opposed to a lab or a shed) that X is 1,000 times better than Y, I will, as they are forced to do (and seem - somewhat curiously - happy to do), have to make do with listening and my own, foolish, subjective opinion.

[written in one go with no proof reading. I may change my opinions at a later date]

Good post, and I tend to agree for the most part.

AB/X is desirable, but inpractical in most situations, other than those that only involve software.

I feel Steve's reply is important too ( mentioning a switchbox).

While AB/X is next to impossible to practically do in a demo, at least being given the chance to blind A/B would be very useful. I wonder how many dealers offer this?

And when it comes down to traditional, passive speaker based systems, I personally feel that most equipment (other than the speakers) are at such a level of competency as to bring the buying decision more down to looks, brand loyalty or attraction, than anything else (I am not saying that there are no differences btw, just that they are smaller these days than at any time in history)
 

fr0g

New member
Jan 7, 2008
445
0
0
Visit site
...and given that AB/X is mostly inpractical, yet AB/X tests are documented all over the internet, I wonder what makes many people simply dismiss the ones that have been done?

I am guessing that unless we personally take part, they don't really have much of a sway on us. Human pride? Is it that we need to "see for ourselves"?
 

CnoEvil

New member
Aug 21, 2009
556
14
0
Visit site
andybeau said:
CnoEvil said:
pauln said:
There are some nice people posting on the headphone section though.

Well excuse us!

They are nice over their Cno

:grin:

I must get me some headphones then.

My last ones were large, black, Pioneer over-ear jobbies, bought in 1977....and now lie unloved (and in bits), in a drawer somewhere.

IIRC they were these ones: http://www.ebay.com/itm/Vintage-Pioneer-SE-205-Stereo-Headphones-Great-Sound-/261212276885?pt=Vintage_Electronics_R2&hash=item3cd1770895
 

busb

Well-known member
Jun 14, 2011
85
8
18,545
Visit site
the record spot said:
pauln said:
fr0g said:
No arguing or nastiness, simply discussion. Is it possible?

Probably not on this forum. Try somewhere more science based. This is predominantly a place for the 'audiophile' not the engineer. You won't be able to have a rational debate. There are some nice people posting on the headphone section though.

The kind of dismissive comment about this place's membership that just serves to wind people up Paul. You're welcome to your views, but there are several qualified people on here. Some even tertiary qualified. Others are regular guys who just want to chat about hifi or technology and others are new to the game. Your problem is? That we're not all scientists or bandy around the phrase "science" enough? Such is life.

I have to agree. Some people use so-called science to bash other people with. The result? They alienate & further polarise many away from rational debate due to arrogance. No one is forced to post here. Don't like it? Vote with your feet. But wait - some have a duty to "truth" & selflessly have to educate the unwashed.
 

CnoEvil

New member
Aug 21, 2009
556
14
0
Visit site
busb said:
But wait - some have a duty to "truth" & selflessly have to educate the unwashed.

I'm sitting here in a mixture of urine vapour, sweat and flies, while agitating my head lice. :twisted:
 

Macspur

Well-known member
May 3, 2010
843
3
18,540
Visit site
CnoEvil said:
busb said:
But wait - some have a duty to "truth" & selflessly have to educate the unwashed.

I'm sitting here in a mixture of urine vapour, sweat and flies, while agitating my head lice. :twisted:

Ah yuk! that's quite put me off my elevenses.

smiley-smile.gif
 

busb

Well-known member
Jun 14, 2011
85
8
18,545
Visit site
fr0g said:
...and given that AB/X is mostly inpractical, yet AB/X tests are documented all over the internet, I wonder what makes many people simply dismiss the ones that have been done?

I am guessing that unless we personally take part, they don't really have much of a sway on us. Human pride? Is it that we need to "see for ourselves"?

An observation - personal experience is very powerful. Who can you trust if you cannot trust yourself? The knack is know when to trust oneself (through experience)!

The problem I have with DB ABX testing is the sneaking suspicion it doesn't work. Not only are they expensive to do properly, they are tedious to take part in (IMO). If they are not conducted with a great deal of care & preparation, many will not believe the results.

/as for ABX tests, I have 2 main concerns:

1. How reliable is short-term memory, even with near instant switching? If it's not reliable, the testing is pointless, surely?

2. The methodolgy is skewed. False positives are averaged out by statisical analysis - no problem there. People go very quiet when challenged over how false negatives should be handled.

An example. Some tests are carried out during a flu outburst. Many of the participants have severe colds that effect their hearing. Most cannot tell the difference between A & B. The conclusion: there ain't no diffeence between A & B! To me, this is not just bad science but very bad science! Until the whole method is shown to handle false responses equally, the resulting conclusions should be viewed with scepticism.

I'd love it if some method proved beyond doubt that all well-made cables sounded the same; that all amplifiers meeting a basic spec all sounded the same because I could then choose on the basis of looks, build quality, expected longevity of the manufacturer etc - all concrete, solid reasons.

Another point is how easily fooled our hearing is. Some people feel very insulted by the whole idea that they can be fooled. They shouldn't be! They just need to ackowledge the fact that their hearing can be fooled rather than is always being fooled. If our hearing was always being fooled, we would know if we were listening to music or a road drill!

Another phrase you will hear in these discussions is Expectation Bias. It does exist - even amongst music lovers! However, it's a term used indiscriminately & does not explain the situations where we expect to hear an "improvement" but don't. An example of expectation bias is when a group of listeners are asked to state their preference of something being tested which they duly do. They are then told that nothing was changed during the test. Many will feel slighted by such "dishonesty" - others learn from such an experience.

If we are willing to acknowledge our hearing is not so damn foolproof, we may just move the discussion on a bit without slashing each other as we all run around in the same circles.
 

busb

Well-known member
Jun 14, 2011
85
8
18,545
Visit site
CnoEvil said:
busb said:
But wait - some have a duty to "truth" & selflessly have to educate the unwashed.

I'm sitting here in a mixture of urine vapour, sweat and flies, while agitating my head lice. :twisted:

Nit!
smiley-wink.gif


I'm using a sharp implement to remove the crust of grime from various parts of my anatomy. oh, my - must remove what feels like putty from my left armpit. But hey - what would you expect from someone who suffers from "Cognative Dissonance"!

:bounce:
 

steve_1979

Well-known member
Jul 14, 2010
231
10
18,795
Visit site
Macspur said:
CnoEvil said:
busb said:
But wait - some have a duty to "truth" & selflessly have to educate the unwashed.

I'm sitting here in a mixture of urine vapour, sweat and flies, while agitating my head lice. :twisted:

Ah yuk! that's quite put me off my elevenses.

smiley-smile.gif

:grin:

I've just read this while eating my breakfast (reheated chicken curry from the Chinese takeaway last night).
 

busb

Well-known member
Jun 14, 2011
85
8
18,545
Visit site
MakkaPakka said:
busb said:
The problem I have with DB ABX testing is the sneaking suspicion it doesn't work.

They seem to work just fine when it comes to speakers.

They also seem to work fine when asking people if they are listening to classical or pop music. My point being they only seem to work when the differences are extremely obvious. Many also argue that they don't work ever - even for speakers - stating that the quoted tests were not very scientific.

Feel free to address my point regarding false negatives though. I maybe entirely wrong & well-conducted DB ABX tests are indeed conclusive. Until someone does a little more than quote examples that validate their PoV without some scientific insight how the tests were conducted, I will remain sceptical.
 

MakkaPakka

New member
May 25, 2013
20
0
0
Visit site
busb said:
My point being they only seem to work when the differences are extremely obvious.

That's the whole point though - some people were claiming there are extremely obvious differences but another group of people disagreed and sought to challenge those claims.

People talk about night and day differences all the time which is at odds with all the test results. Actual test results are showing that any differences are very minor.
 

steve_1979

Well-known member
Jul 14, 2010
231
10
18,795
Visit site
busb said:
1. How reliable is short-term memory, even with near instant switching? If it's not reliable, the testing is pointless, surely?

While short term auditory memory is usually pretty good (especially with repetition) I have to also agree that it isn't perfect. However I do think that short term memory it's much better than long term memory. When it comes to makng A/B comparisons there are four options and all of them have pros and cons and non of them are perfect.

1. Making comparisons in any sighted tests. (IMO this can potentially be unreliable due to expectation bias)

2. Using your long term memory in blind comparisons and listening over long periods of time before swapping. (IMO this has okay reliabilty but will often lead to mistakes too)

3. Using your short term memory with an instantaneous A/B switching method. (IMO this has much better reliability but mistakes can still be made)

4. Using equipment to take scientific measurements. (IMO this is the best and most reliable method but only if the correct and relevant measurements are taken accurately and even then someone who is qualified is often required to correctly interprit the results without mistakes being made)

busb said:
2. The methodolgy is skewed. False positives are averaged out by statisical analysis - no problem there. People go very quiet when challenged over how false negatives should be handled.

In ABX tests if people just randomly think that they hear a difference when there isn't really one there then the results should come back almost exactly 50% correct and 50% incorrect. This is very straight forward IMO.
 

DocG

Well-known member
May 1, 2012
54
4
18,545
Visit site
steve_1979 said:
busb said:
2. The methodolgy is skewed. False positives are averaged out by statisical analysis - no problem there. People go very quiet when challenged over how false negatives should be handled.

In ABX tests if people just randomly think that they hear a difference when there isn't really one there then the results should come back almost exactly 50% correct and 50% incorrect. This is very straight forward IMO.

OK, to put busb's remark somewhat sharper (may I, busb?): what if the participants all suffered neurosensory deafness ànd were really honest? Not a single positive result would be reported, so there is no difference, right? Utter :hs:, of course! I'm aware this is a daft example; it's just to make the point clear. :)
 

steve_1979

Well-known member
Jul 14, 2010
231
10
18,795
Visit site
DocG said:
OK, to put busb's remark somewhat sharper (may I, busb?): what if the participants all suffered neurosensory deafness ànd were really honest? Not a single positive result would be reported, so there is no difference, right? Utter :hs:, of course! I'm aware this is a daft example; it's just to make the point clear. :)

This is a very good point that you and busb make. :) ABX testing by definition has one very big limitation - it can only prove a positive result but it can't prove a negative one.

Basically if you really can hear a difference then you will get a positive result which categorically proves that you heard a difference. That's all fine and dandy and provides good quality useful data. However the problem with ABX testing is that you can't prove a negative result. If you can't hear any difference it doesn't prove that a difference can't be heard. It just shows that those people on that particular occasion didn't hear a difference. If they tried it again another time they might be able to hear a difference.

I fully agree with you that ABX testing isn't perfect. But it's still the most reliable comparison method available for checking whether or not you can hear any differences between two things. Don't forget that it's not just ABX which can't prove a negative result but all other listening comparison methods can't either.
 

fr0g

New member
Jan 7, 2008
445
0
0
Visit site
I think one point is being missed here.

A single ABX test cn only prove a difference, not a lack of one. If we have a large enough pool of participants then it cn go some way to prove a lack of one.

And in either case, "prove" is not 100%, it's statistical probabilities to a smaller or larger extent.
 

TRENDING THREADS

Latest posts