subjective listening tests

For the last few months i have been recapping a TAC Scorpion II, 

Going forward, i would like to do some subjective op-amp experiments.

Usually if i want "hear" the difference between 2 very similar audio clips, i will mix both clips against each other, with the 2nd clip,  out of phase.

Would this be a good way to compare op-amps differences?

Is there a better way to do this?

pb said:

For the last few months i have been recapping a TAC Scorpion II, 

Going forward, i would like to do some subjective op-amp experiments.

Usually if i want "hear" the difference between 2 very similar audio clips, i will mix both clips against each other, with the 2nd clip,  out of phase.

Would this be a good way to compare op-amps differences?

Is there a better way to do this?

Click to expand...

The problem with null-tests is that:
A) you need to make sure levels and frequency response are matched to less than 0.01%, which IMO is impossible with analog gear.
B) You may identify differences, but you will never know which one is "more different"; they're supposed to be equal, but one is probably "more equal"*, or not...?
If you hear HF in the difference signal, what will tell you if it's channel 1 that has too much treble or channel 2 that has not enough, or one has too much and the other not enough...
I've never used null-tests for these reasons; I'm also very wary of blind tests, which unfortunately I have to do regularly in the course of my professional activity. It takes only a few seconds to get accustomed to a sound and consider it is the actual reference; practicing blind tests actually requires serious training in analysing the different aspects, in a somewhat similar fashion as a wine-taster analyses separately the elements of taste.
I'm a firm advocate of objective tests, using measurement instrumentation, and relating them to subjective evaluation.
Indeed, that applies to linear processing, such as amplification, EQ, mixing or compression; when it comes to FX, the ear is the only judge. What's the point of measuring a TS9 against one of its numerous clones?
* Animal farm (George Orwell)

Abbey, thank you for the thoughtful response.

As it is, i am very happy with how this desk is sounding, and it will probably stay stock.

abbey road d enfer said:

I'm a firm advocate of objective tests, using measurement instrumentation, and relating them to subjective evaluation.

Click to expand...

I agree.

I am with Abbey on the value of bench testing audio paths but null testing is an easy way to prove there is no difference while difficult to get a deep null when even a little phase shift or frequency response difference exists between the stems being nulled.

I have suggested null testing for comparing console channels to each other to parse out an outlier bad channel or maybe marginal deteriorating capacitors.

Since 99.99% of op amp circuits use negative feedback, the transfer function (output vs. input) will be dominated by the components used in the NF path.  Op amp parameters can make a difference in the margin. I expect high noise gain stages to show the most sensitivity but wouldn't expect much difference if the original design is well executed.

JR 

pb said:

Going forward, i would like to do some subjective op-amp experiments.

Usually if i want "hear" the difference between 2 very similar audio clips, i will mix both clips against each other, with the 2nd clip,  out of phase.

Would this be a good way to compare op-amps differences?

Click to expand...

If you are interested in the reason why you hear differences, have a look at ...

Kingston's excellent thread on what is REALLY important for noise & THD; opamps and local decoupling of rails, some questions

http://www.groupdiy.com/index.php?topic=37307.80

Many true gurus chime in.  It proves how OPA rolling takes a VERY poor second place to correct earthing, layout & decoupling.  It’s a long thread but read the whole thing from #41 to find pearls of wisdom.

As a true DBLT guru for nearly 2 decades in my previous life, I have to disagree with Abbey about the need for 'serious training' in DBLT.  If the test is configured properly, the 'man in the street' usually outperforms (gives more consistent results than) nearly all HiFi Reviewers.

But its VERY expensive to do DBLTs properly. eg the victim picks his own music and has complete control over switching and levels.  One presentation may take 1/2 day and that's testing 1 person.

Having said that, I agree with him & JR about "measuring and relating these to subjective evaluation"

I went through refurbishing and modding a TAC Scorpion some years ago. IIRC I already posted  several findings in a thread here, but let me explain a few things again. Modding for the sake of throwing money into the desk and hoping things will improve makes no sense. It makes totally sense, though, to improve certain parameters in a design. Verify through measuring!

Scorpions were made as a budget desk, so the designers had to do quite some design compromises:

The pots used have high tolerances causing serious level differences between channels. There´s no way to trim levels. So you will run into level problems when comparing channels. Be aware of that.

The micpre chip (SSM2015) is not good in driving loads. In a perfect world it would have a buffer following it. In this desk it doesn´t. So when you don´t engage the EQ the SSM chip has to drive the insert. Hence it´s not a good idea to connect 600 Ohm gear to the insert. If the EQ is on then the SSM has to drive the EQ which is ok loadwise, but then the LF-band chip (TL072) of the EQ has to drive the insert. Again not the best choice for 600 Ohm gear because TL072 don´t like loads <2kOhm. If you want to improve the design then swap this chip for something capable of driving 600Ohm. OPA2132 or OPA2134 or TLE2072 are good chips for that purpose. All of them have better noise and distortion behaviour as an additional bonus.
If you want to swap chips to faster ones then you generally have to add local supply bypass caps as close as possible to the chip´s B+/B- input pins, at least closer than 1cm to ground. For stability 0,1uF ceramics are fine. If you want to improve transient response then more local decoupling is necessary. 10uF per amp is a good value. But these are very hard to fit into the Scorpion´s PCB design. That´s a challenge. But go for it, it will help a lot!
IIRC the design team was very good in adding bandwith limiting caps to every chip. So chances are low that you will run into oscillation problems. But it´s a good idea to check for oscillations after every single step you did.

I highly recommend to change the local supply decoupling at the micpre chip. There are 0,01uF ceramics from rails to ground close to the chip. Remove them and install 100uF for each rail instead + add the original 0,01uF ceramics parallel to the electrolytics underneath the PCB.
These desks were made in times of tape and shitty PA systems. Back then there was not much going on below 50 Hz. So the LF corner of that design might be a little high by todays standards. When you want to improve that then increase all coupling caps by 2x - 4x. This will lower phaseshift in the lowend hence the lows will be more controlled, go further down and the lowmid range will sound cleaner due to the removed phaseshift.
The summing amps are TL072 again. Not the best choice because you want higher open loop gain here.  There are many better chips for this purpose. In my desk there are OPA2134 in this spot now. But read datasheets to find other alternatives. Don´t forget local decoupling!
Outputs are a pain because they are unbalanced and even worse with some funny pinout for pro and consumer level signals. It´d make sense to change that. Most or even all jack outputs can be changed to impedance balanced by removing one resistor and changing another one on the output PCBs. Working on these is a mechanical nightmare, though.
The ribbon wires and IDC connectors from I/O boards and to the faders have selfcutting contacts and are prone to oxidation. To make your desk longtime reliable throw all that out and put new ones in. Same applies for the ribbon wires between master and group PCBs.
The fader amps are TLo72 again. They don´t work in their best operating point noisewise in these spots. BJT input chips may be better suited here. IIRC I have installed LT1358 here, but it could be something else either. LT1358 are superduperfast chips, so don´t forget local decoupling.
Oh, and did I mention that local decoupling helps a lot?

Back when I was doing all that to my desk I did use my collegue at work for blindtesting and evaluating all these mods. If he couldn´t hear them I didn´t install them in my final version. At some point he was wading through 12 differently modded channels, taking notes of all the differences he was hearing. That gave me in the end a lot of confidence that I didn´t fall into the trap of subjectivism.

I hope all that makes sense and helps.

Auditory Research tells us that about 20% of any randomly selected group have a consistent and repeatable ability to identify two sonic signatures where they determine that they differ, and about 10% are able to correctly discern that  they are identical.

Other research shows that a majority of ordinary listeners prefer a sound sample with predominantly high order harmonic distortion over one with predominantly low order HD initially, but after approximately one hour of listening they reverse their preferences.

Very little of the current literature employs actual "High Fidelity" audio equipment. For example they may use headphones with an upper frequency response limit of 8 KHz, because that's the lab spec equipment audiologists use, and research must be repeatable in order to be taken seriously; it all has to be peer reviewable.

More than half of all scientific or medical research findings cannot be repeated and therefore the conclusions are declared invalid by the Scientific Method; peer review is a very big deal so researchers don't like to mess with it.

It doesn't make conducting your own listening tests easy.

Good luck. But please trust your own conclusions; you might be one of the 20/10%.

I don't find null tests all that useful,  except if trying to see if things are identical.

Objective test and measurement is very useful to quantify differences.  But it doesn't tell you how something sounds.

I also like subjective testing.  Record the same source material through different circuits, match the levels,  and then listen. I also record the circuit in true bypass to hear what the converters are doing. This is not a true scientific test,  but I find very useful to identify what I find socially pleasing.

jensenmann said:

I went through refurbishing and modding a TAC Scorpion some years ago.

Click to expand...

Congratulations for the thorough and detailed approach you handled this refurb with!

The fader amps are TLo72 again. They don´t work in their best operating point noisewise in these spots. BJT input chips may be better suited here.

Click to expand...

I've been there before; there is a compromise to be found between the "static" noise (almost entirely due to the opamp's noise-voltage and noise-current) and the "dynamic noise that results from the opamp seeing a changing source impedance as the fader moves. With a TL072, there is only "static" noise, because FET-input opamps have negligible noise current; replacing them with 5532's give a much lower "static" noise (about 10 dB less) but rapid fader movements become audible because of noise modulation. I would tend to favour low noise FET opamps, such as OPA2134 (about 7dB improvement).

I hope all that makes sense and helps.

Click to expand...

Yes it does. Thanks!

Johnny2Bad said:

Auditory Research tells us that about 20% of any randomly selected group have a consistent and repeatable ability to identify two sonic signatures where they determine that they differ, and about 10% are able to correctly discern that  they are identical.

Click to expand...

Have you got links to this Auditory Research?

My nearly 2 decades of DBLTs suggests those who can correctly discern that they are identical is very small.  They are the true golden pinnae .. unlike the wannabe Golden Pinnae who will say "Chalk & Cheese" if you present the same item twice.

ricardo said:

As a true DBLT guru for nearly 2 decades in my previous life, I have to disagree with Abbey about the need for 'serious training' in DBLT.  If the test is configured properly, the 'man in the street' usually outperforms (gives more consistent results than) nearly all HiFi Reviewers.

Click to expand...

Although I agree that the "man in the street" is capable of giving a consistent opinion without training (meaning he can repeatedly identify which one is which and giving a subjective evaluation), I don't think he is capable of a detailed evaluation. I mean most of the times, he will prefer the system that has more bass and treble, less nasal midrange, favor 2nd-order harmonic distortion, and so on, but will not be capable of isolating these characters.
Trained "listeners" know how to isolate particular areas of the system's reproduction, and evaluate them independantly, like a wine-taster can evaluate separately the sugar, alcohol and tanine contents, the different aromas and constituents of sapidity.
An untrained listener will tell you "I like this system better than that one", a trained one will tell you " I lke the bass, but not so much the upper midrange, and I think there's some distortion here"; to me the latter is more useful, because it allows me to concentrate on a specific area, instead of sitting here pulling my hair.
When the differences between two systems are multi-dimensional (different frequency response, different distorsion behaviour, different recovery time, different crest factor), an untrained listener may mark as equally good (equally bad?) two flawed systems, without being capable of pinpointing the high and low points.
The simple "I like it better" comment may be conflicting with the design target, particularly if you're targetting at "truest".
I believe you are very familiar with the notion of acquired taste. I think you can't prevent that notion to interfere with all subjective tests. Part of the training for listeners is to be capable of isolating from this notion (not ignoring it, which is impossible IMO) by decomposing a global performance in separate factors. Some are naturally capable of doing it to a point instinctively, but it's a gift that needs to be nurtured.

As to "HiFi Reviewers", I recently went to the doctor's and read a review of digital music servers (no one would actually catch me buying an audiophool magazine). I had the pleasure to know that some were notable for their "suavity"  :

abbey road d enfer said:

Although I agree that the "man in the street" is capable of giving a consistent opinion without training (meaning he can repeatedly identify which one is which and giving a subjective evaluation), I don't think he is capable of a detailed evaluation. I mean most of the times, he will prefer the system that has more bass and treble, less nasal midrange, favor 2nd-order harmonic distortion, and so on, but will not be capable of isolating these characters.

Click to expand...

Actually the result of nearly 2 decades of DBLTs show that ...

.. the man in the street, the woman who disclaims any interest in HiFi right down to the recording engineer who makes his own mikes and insists on using his own recordings for assessment ...

actually 'like' the same speakers (most of my DBLTs were on speakers though there were important studies on electrical & digital stuff too) IF the test is configured properly eg the victim chooses his own music bla bla.

This was a surprising result as we expected the headbanging pop fan to have different tastes to the classical recording engineer etc.  Hence I believe there is such a thing as 'good' speakers which are 'liked' by everyone in DBLTs and I try to design speakers & other stuff that meet this as a priority.

We have tried loadsa complicated forms etc in the 2 decades but the easiest & most useful was a simple score out of 10 for each 'presentation'.  The listener was encouraged to make comments and this is always sufficient to identify if there is too much bass etc.  We insist on at least 5 pieces of music.  There is no upper limit.  But they have to include

• something simple with vocals
• something complicated with loadsa instruments and/or stuff happening
• something loud
• something quiet

The best speaker out of those 2 decades was a little 6 ltr box which came out top in more than a dozen DBLT test series .. usually against larger & much more expensive speakers.  No other speaker has even come close to this record so I feel justified in claiming it is probably (still) the best small speaker ever made.

It had a true bass extension of 70Hz but consistently garners praise for 'musical' bass.  Only a couple of recording engineers have commented that it probably didn't have extension below 50Hz but they still said the bass was very well balanced.

It was one of our products and I'm told there is a lot of me in it.  But apart from the bass, which I think I can replicate, I can't put my hand on my heart and say these are the features which made it sound good as we had other speakers with  the same features. 

Of course when you do DBLTs, you rapidly find all audiophools and nearly all reviewers are deaf. 

Because they give inconsistent results, there is no point trying to design for a sound they like as if you put the same thing in twice, they will give radically different opinions.  BTW, we did ABC rather than ABX tests as you get statistical significance a lot quicker.  I think there is now an AES recommendation for ABC tests.

The strategy for these wannabe Golden Pinnae is to repeatedly say loudly and clearly that your stuff is hand carved by Virgins from solid Unobtainium.  A big $$$ tag helps too. 8)

abbey road d enfer said:

Although I agree that the "man in the street" is capable of giving a consistent opinion without training (meaning he can repeatedly identify which one is which and giving a subjective evaluation), I don't think he is capable of a detailed evaluation. I mean most of the times, he will prefer the system that has more bass and treble, less nasal midrange, favor 2nd-order harmonic distortion, and so on, but will not be capable of isolating these characters.
Trained "listeners" know how to isolate particular areas of the system's reproduction, and evaluate them independantly, like a wine-taster can evaluate separately the sugar, alcohol and tanine contents, the different aromas and constituents of sapidity.
An untrained listener will tell you "I like this system better than that one", a trained one will tell you " I lke the bass, but not so much the upper midrange, and I think there's some distortion here"; to me the latter is more useful, because it allows me to concentrate on a specific area, instead of sitting here pulling my hair.
When the differences between two systems are multi-dimensional (different frequency response, different distorsion behaviour, different recovery time, different crest factor), an untrained listener may mark as equally good (equally bad?) two flawed systems, without being capable of pinpointing the high and low points.
The simple "I like it better" comment may be conflicting with the design target, particularly if you're targetting at "truest".
I believe you are very familiar with the notion of acquired taste. I think you can't prevent that notion to interfere with all subjective tests. Part of the training for listeners is to be capable of isolating from this notion (not ignoring it, which is impossible IMO) by decomposing a global performance in separate factors. Some are naturally capable of doing it to a point instinctively, but it's a gift that needs to be nurtured.

As to "HiFi Reviewers", I recently went to the doctor's and read a review of digital music servers (no one would actually catch me buying an audiophool magazine). I had the pleasure to know that some were notable for their "suavity"  :

Click to expand...

I spent way too many hours trying to undo damage from improperly controlled listening tests and barely trust my own ears to tell more than if the needle is dirty or clean (remember that?).

I have told this story before.... back last century when Peavey was pimping their solid state tube emulation amps they did a single blind A/B test in the sound booth between a real tube amp, and the solid state mimic. The participants were allowed to play guitar through the two amps and switch back and forth as many times as they wanted.

The test involved music store owners and salespeople who should have a some knowledge of what tube amps sound like. The vast majority could not pick out the tube amp with any statistical significance. That said a tiny handful of participants, typically professional musicians, were able to pick the real tube amp every time. They were apparently able to tune into some subtle sonic cues that were different between them.

So to reinforce what Abbey is saying, trained or skilled listeners can hear things that unskilled listeners can't.  Back when I spent so much bench time tweaking dynamics designs (mostly for companding NR) I became very sensitive to side chain caused artifacts, and often hear stuff in TV broadcast audio feeds that shouldn't be there.  :'(

I trust unsophisticated listeners to pick what they like, but maybe not pick the same one next week.

I prefer to deliver a flat/linear audio path and let the source material determine the winners and losers. The free market will determine the source material winners and losers (in an ideal world).

JR 

Great replies, & links to some very informative documents, really appreciated here.

Spent the weekend, finishing a half normalised, re-jigged 96 point TT patchbay.

I had been psyching myself up, for doing this, since before Christmas.

The main mental block for me to getting started, was actually ripping the old configuration down.

but now



I just need to pick up a couple of extra Edac chassis sockets (for breakout), & fit them, then this loom will be done.

jensenmann said:

The pots used have high tolerances causing serious level differences between channels. There´s no way to trim levels. So you will run into level problems when comparing channels. Be aware of that.

Click to expand...

Jensenmann,

Over the weekend, i was feeding a 0dB sine signal, from my sound card, to the desk channels one at a time, all channel gains were set at FCC minimum, all channel faders were set to zero, the most I had to move a fader, to match to 0dB, was 2mm travel, on one channel, & the rest were within a gnats dick width of being true. (30 Inputs)

I have not lifted and serviced the fader panel yet, still on the "to do" list, so the faders might sharpen up a little bit yet, with some TLC.

Hopefully will get onto the eq's later in the week.

best

pb

ricardo said:

Actually the result of nearly 2 decades of DBLTs show that ...

.. the man in the street, the woman who disclaims any interest in HiFi right down to the recording engineer who makes his own mikes and insists on using his own recordings for assessment ...

actually 'like' the same speakers (most of my DBLTs were on speakers though there were important studies on electrical & digital stuff too) IF the test is configured properly eg the victim chooses his own music bla bla.

Click to expand...

I believe your view and mine are not in disagreement at all, but have different purposes.
You use Subjective Listening Tests for confirmation whether a product is good or not, and even grading; I use SLT for analysing the areas that need adressing and those that don't.

It was one of our products and I'm told there is a lot of me in it.  But apart from the bass, which I think I can replicate, I can't put my hand on my heart and say these are the features which made it sound good as we had other speakers with  the same features. 

Click to expand...

You admit that these SLT were not really useful to help you designing the product; it was a confirmation that yours was good and better than others. What would have you done if your speaker was not evaluated as a good one?

Of course when you do DBLTs, you rapidly find all audiophools and nearly all reviewers are deaf. 

Click to expand...

It's not so much their inabilities that bother me, it's the pedantic tone and pompous language.

BTW, we did ABC rather than ABX tests as you get statistical significance a lot quicker.  I think there is now an AES recommendation for ABC tests.

Click to expand...

Huh? My understanding was that ABC involved a permanent reference, a concept that is understandable in the context of evaluation of transmission, but definitely unappropriate for transducers. We're probably talking of different things...

pb said:

Great replies, & links to some very informative documents, really appreciated here.

Click to expand...

Congrats for keeping focused on the original subject, while we hijack your thread!

I spent a lot of time listening to op amps and optimizing grounding, decoupling etc. in my console as well as other gear. Even with the optimizations the op amps all have an identifieable sound. I always come back to the TI NE5534/NE5532 as the least objectionable in most situations (and as long as it's properly implimented).

For tests I like to let the components warm up, record, replace the op amps, warm up again and record again. Nulling recordings of op amps against themselves and other op amp types confirms differences and similarities. 

abbey road d enfer said:

You admit that these SLT were not really useful to help you designing the product; it was a confirmation that yours was good and better than others. What would have you done if your speaker was not evaluated as a good one?

Click to expand...

We do a lot of Subjective Listening Tests during the design of a product often against other product, ours and other makers, with similar design constraints.  These are always done Double Blind.  Price isn't a factor in this.  For a speaker, the box size is the usual criteria.

The design might change during this process to ensure it beats competing product.  Once the design is finalised, it might be tested against much bigger and expensive stuff.

This particular speaker was exceptional because of the final product's record of coming top in every single DBLT it was in.

Before it, we (the R&D Dept) counted 3 speakers as the 3 best small speakers you could buy because they had about equal scores in various DBLTs.  The cheapest was my design and was the 2nd best selling speaker in Europe for at least 7 yrs.  The 2 from other makers were more expensive.  One MUCH more expensive which also needed at least 200W to perform well in DBLTs.  The 3rd had about the same 'score' but was less consistent than my design .. usually generating love or hate comment.

So we thought we had the best balance of qualities. 

So yes.  We do use Subjective Listening Tests to design stuff but they have to be DBLTs.  I suppose that's one of my true guru credentials in my previous life.

My design was really cheap and looked it.  I consistently give it higher scores Blind than when I know I'm listening to it.

You can easily bias DBLTs once you get some experience.  eg using a 50W amplifier would significantly downgrade the expensive small speaker.

But its so expensive to do DBLTs properly that you never waste them on trivial marketing stunts.

Huh? My understanding was that ABC involved a permanent reference, a concept that is understandable in the context of evaluation of transmission, but definitely unappropriate for transducers.

Click to expand...

It's the ABX test that assumes an 'identified' reference.  There is a 50% chance of a 'correct' result if the victim is deaf.

In a proper ABC test, the victim is NEVER told what he is listening to.  It might be 3 speakers or it might be 2 different types of speaker cable, one of which is repeated .. or something like dither tests.

A DBLT is a MEASUREMENT.  Your instrument is your DBLT panel.  It has an accuracy that you need to check & calibrate regularly.

You do that by putting the same item  in 2 of A, B or C.  There is only a 33% chance of a 'correct' result if the victim is guessing.  If the victim does the test 3x and comes out with the same result, you can put a LOT of confidence in his opinions..

You discount the opinions of anyone who gets the 'wrong' answer .. usually the audiophools who will ascribe chalk & cheese to 2 presentations of the same thing.

I apologise to da statisticians reading this for this foray into 'limited' sampling.  8)

ricardo said:

This particular speaker was exceptional because of the final product's record of coming top in every single DBLT it was in.

Click to expand...

I think you are taking a shortcut here. I would think, this speaker was exceptional because any problems detected during the design phase was ironed out. Detecting these problems by DBLTing them is just one of the possibilities. The fact that it won all the DBLT's is a consequence, not a cause. And again, you can't make a DBLT out of thin air, you must have a product to try. The design does not come from DBLT's, it comes from setting a design brief and working the compromises (or maybe from pure serendipity). Now, indeed, final approval must be corroborated by subjective tests.

So yes.  We do use Subjective Listening Tests to design stuff but they have to be DBLTs.  I suppose that's one of my true guru credentials in my previous life.

Click to expand...

I still don't understand how you can expect the results of a DBLT to help in identifying specific issues like cone break-up or directivity anomalies, unless you have attracted the attention of the listeners to these particular issues, which I would qualify as "training".
My experience with untrained listeners is that as you move the fader up, the sound becomes better...

It's the ABX test that assumes an 'identified' reference.  There is a 50% chance of a 'correct' result if the victim is deaf.

In a proper ABC test, the victim is NEVER told what he is listening to.  It might be 3 speakers or it might be 2 different types of speaker cable, one of which is repeated .. or something like dither tests.

Click to expand...

Unfortunately my memory fails me on the subject and I found that googling ABC returns anything from basket-ball to alphabet to hepatitis. The closest I got was ABC-hr (hidden reference). I don't know how that would apply in speaker testing, since it involves the existence of an undisputed reference.

abbey road d enfer said:

I still don't understand how you can expect the results of a DBLT to help in identifying specific issues like cone break-up or directivity anomalies, unless you have attracted the attention of the listeners to these particular issues, which I would qualify as "training".

Click to expand...

We were among the first to do serious work on cone breakup with laser doppler interferometer bla bla and look at directivity over a large frequency range.  Remember we designed and made units including cones, surrounds etc. to go in our speakers.

This all comes under 'objective' measurements which are verified by DBLTs.  The reason for wanting well controlled cone break up, nice directivity etc is cos ... 'our' prejudice from conducting DBLTs is that it leads to speakers that most listeners, from recording engineer to woman in the street, 'like'.  These are also, perhaps surprisingly, accurate speakers.

The subject of 'nice' cone breakup is another subject on which I can pontificate forever.  eg its different for a paper cone compared to a cone made from an engineered plastic.

Unfortunately my memory fails me on the subject and I found that googling ABC returns anything from basket-ball to alphabet to hepatitis. The closest I got was ABC-hr (hidden reference). I don't know how that would apply in speaker testing, since it involves the existence of an undisputed reference.

Click to expand...

No.  An ABC test just asks the victim to rate 3 presentations.  No other information is offered. 

An ABX test asks the victim, "which of A or B is closest to X?"  The X is the 'undisputed' reference.

The ABC test can be used to obtain the null result ("there is no audible difference") too ... but much more quickly to achieve a specified level of statistical significance.  I suppose if the test is eg to determine the audibility of phase distortion, the undistorted signal path is the 'hidden reference' .. but it could be repeated.  BTW, this was an important DBLT test series for us.

Distortion in speakers is difficult.  I had several cases of speakers with higher measured THD do better than speakers with much lower distortion ... with comments that these seemed very undistorted with a very experienced & perceptive panel.  The important audible distortions in speakers are gross ... but not what most people measure.

Look up AES papers by Fryer for some of our early work on this.

BTW, though I've pimped the 'man in the street' as being more perceptive than most (all?) audiophools & HiFi reviewers, the best (most consistent) people on my DBLT panel were some loudspeaker designers & a couple of recording engineers .. though there are important deaf buggers even in this august company.

Only 1 HiFi reviewer is in this A team .. out of all UK & US reviewers.