Interview with Dynaudio chief engineer Stephen Entwistle

[lowkeyoperations]

Doesn't it all depend on the scale of the variations ?
As @MattHooper said, you can easily distinguish between your relative's voices over the phone without any hesitation even weeks after not having listened to them.

I would keep open the possibility that it is about more than the scale of the variations.

We each know hundreds of voices that we can distinguish and recall, whether they are people we know, radio presenters, actors etc.
Sometimes I remember the voice better than the name.

Overall, the human voice sits in a fairy narrow band of frequencies. How different do human voices measure in FR? I don’t know. But given that there are hundreds of people I personally know, I dont think I’ve had the experience of getting their voice wrong on the phone.

So I’d say that differences in voice could probably be defined as fairly subtle.

Perhaps other factors beyond the scale of variation come in to play. Like repetition. I’ve heard my wife’s voice a lot. I could identify that voice very easily. She’s away on holidays for a month. But I won’t forget it. It is literally etched into my brain.

On the other hand, I can’t say that I would remember the voice of a shop assistant who I spoke to for the first time just a few minutes ago. Possibly because a- it’s the first time I heard their voice and possibly because b-the interaction was ‘meaningless’.

So repetition and meaning are both potentially part of what forms memory, as well as scale of variation.

This makes sense to me, as I’m a musician, hifi enthusiast and audio producer.

When listening to speakers for the purposes of assessment, it’s pretty standard practice to listen to a collection of reference songs. Why? Meaning and repetition I’d say. I would say I can remember the stereo spread on the words “Another one bites the dust” where the vocal double effect is placed, because I have heard it literally hundreds of times on scores of different speakers. I’ve studied it in detail on my studio monitors to tried to reproduce it in some productions. It’s just one example of audio memory that like my wife’s voice is etched into my brain.

I know a couple of my reference songs are borderline sibilant. I know some have a little reverb tail on a sound that is audible to varying degrees on each speaker. Some have deep bass lines that may or may not disappear somewhat.

Listening to a collection of reference music on a pair of speakers in your own house for a couple of weeks gives you the time to re-experience those memories and make an assessment. You can also do it in a ‘less focused’ and ‘more relaxed’ manner that is potentially closer to how you would listen to music for enjoyment as opposed to listening for difference. (As a music producer you have to focus on both minute differences and overall ‘vibe’ to make a successful mix).

Switching backwards and forwards quickly gives you a better ability in differentiating the character of a sound. But listening to the whole song in context gives you the ability to judge overall if the decision was right.

Switching backwards and forwards might help tell you which speaker is more revealing in a certain frequency. But I think extended listening will help you determine which of those differences give you more enjoyment overall when listening to music.

So think there are different types of listen too.

In my studio, I spend a higher percentage of my time actively listening for difference. The speakers I use in the studio are very good for that, and their nearfield setup benefits that use case. They aren’t that great for purely ‘enjoyment’ listening.

In my lounge I spend a higher percentage of time listening for pleasure. I’m not wanting to lay on the sofa constantly analysing to sound. I want to relax. I want to hear something that sounds right to me.

So who knows what Dynaudio were listening for in their tests, and how it was undertaken? I don’t think the interview outlined all of that.

But in my experience, taking a speaker home for a couple of weeks to listen to music I know, is a beneficial process in assessing how much I enjoy listening to them. And for many of us here, the deep enjoyment of music is the ultimate goal.

Sorry for the long, rambling post based only on subjective experience and memory. that’s my coffee finished.

 

[markus]

Typical audiophile confusion about creating sound reproduction devices, creating recordings and creating an enjoyable listening experience. This thread is about the former.
Building sound reproduction devices (and evaluating their performance) is an engineering task. And as such it has to rely on measurements, i.e. a "general obsession with measurements" is just the nature of the task. The human ear is not a good nor effective measuring instrument. There are ways to use it as such but it involves elaborate blind testing protocols.
Are we measuring all relevant performance criteria yet? Probably not but that's not an excuse to defend the highly flawed process of sighted listening. By doing so you just shoot yourself in the foot by cementing the status quo.

 

[MattHooper]

Typical audiophile confusion about creating sound reproduction devices, creating recordings and creating an enjoyable listening experience.

It's no confusion to put claims like "our auditory memory is pretty bad" and "sighted listening is just a massive waste of time" in to some context.

That's actually untangling confusion.

This thread is about the former.

You responded to Tuga who explained his own preferred method of evaluating speakers, involving longer term listening to each.

You disagreed with the validity of his personal method, arguing it would "hardly be useful" to learn something objective about the real world, citing sighted bias and that our "auditory memory is pretty bad." You concluded with "sighted listening is really just a massive waste of time" if someone wants to learn something real ("Objective") about the gear.

So you were not simply aiming at speaker development and design; you were also directly taking the method of individual audiophiles to task.

This is why I asked you to clear up what seemed to be some overreach and muddying of the waters on your part. Look for instance at my replies to "abdo123" concerning the research on auditory memory. It showed our auditory memory is LESS reliable (in the sort term!) vs sighted memory; it didn't show auditory memory is "simply unreliable!" It actually showed it was fairly accurate! Enough to be useful (which is why it evolved in the first place!) As I said: anyone starting with the claim that our auditory memory is "bad" and "unreliable" has to be able to put that in context and explain how I could be doing my job as a sound editor sucessfully, day to day. You can't just "reason in a bubble" ignoring the implications of things you argue.

Remember that even science doesn't deliver Absolute Certainty and rarely 100 percent confidence levels, so we are never talking about Absolute Known Truth per se, but always dealing with confidence levels: all things considered is it REASONABLE to believe X?

Combine that fact with the other fact that almost all of our daily experiences and inferences are made WITHOUT scientific controls, means that there will be a large grey area for confidence levels. Pragmatism (an important element of epistemology) will council us to provisionally accept all sorts of conclusions as reasonable...despite not being under scientific controls.

This can apply to purchasing and evaluating audio gear, as to anything else we do.

So, back to Tuga's method of evaluating speakers, in which he wants to assess speakers by listening to them long term in his own room.
Will this deliver a SCIENTIFIC level of confidence, without scientific controls? No.

Does that entail it is therefore a "massive waste of time" and an unreasonable method of evaluating speakers? Surely it isn't. Our acoustic memory isn't perfect, but it's not useless! Our sighted evaluations aren't perfect, but they aren't necessarily useless! Pragmatically speaking most people can't employ peer-reviewed scientific controls when auditioning speakers. This is the gray area where pragmatism enters.

So, while there are liabilities involved, you can't say Tuga's method WILL NOT allow him identify any "truths" about the speakers. It may well do so. Insofar as a speaker has audible differences compared to his previous speakers it's entirely plausible he may pick up on those differences (e.g. if the new speaker is too bright, or it's inciting bass nodes in his room the other speakers did not, etc). To choose a more obvious example, if Tuga is used to listening to music on tiny, little LS3/5s and is now auditioning full-range Revel Salon 2s in their place, is it unreasonable to think he won't require scientific controls to identify actual sonic differences when playing a range of music? Is it unreasonable that he could, over time, discern which speaker is more satisfying to live with when playing a range of his favorite music?
If he perceives the Revel Salons bring the fuller grandeur of the scale and dynamics of an orchestral piece in a more satisfying way over the LS3/5s, isn't it entirely plausible, even likely, he's perceiving REAL sonic differences?

In other words, let's keep some perspective here.

 

[MattHooper]

It's interesting that some of the same cautions seem required here sometimes that come up in the more subjectivist forums.

Take a common "subjectivist" argument "But we rely on our senses all day long and they work! So therefore I can rely on my senses to tell me the truth about how AC cables sound different!"

What's the problem there?

It's that the principle they are appealing to in their argument - the "reliability of our senses" - actually ignores all the data we have on when our perceptions are UNRELIABLE. The way bias can confound our perception. The limitations of auditory memory etc.

This is why I am so often pointing out to the "subjectivists" that they can't just reason as if the claims they make are fully detached from the implications outside the audio gear realm. Audio gear doesn't exist in it's own bubble, detached from what we know about engineering, and detached from what we know scientifically about the problems of sighted bias. If they were looking at the coherence of their arguments in the big picture, they wouldn't be making those type of mistakes. They'd say "actually, engineering theory and practice suggests that properly designed AC cables are unlikely to make audible differences, and if I seem to hear otherwise I'm going to want to be extra cautious and in a listening evaluation, employ controls for bias!"

The same goes for when an "objectivist" is arguing within such a narrow scope that he's not really paying attention to the wider implications of his argument. Just as the subjectivist who argues for the reliability of our senses needs to put that in context of the known areas of unreliability, the objectivist appealing to the "unreliability of sighted listening and acoustic memory" has to be able to put that in the context of the the known reliability of acoustic memory, and the fact we seem to successfully navigate much of the world, on a daily basis, without scientific controls.

It's why I often appeal to how work gets done in my own field day in and day out, despite the lack of scientific controls. When some people respond that these are "two different things - the assessment of audio gear vs creating content" - this is a confusion. It's a confusion like the subjectivist arguing that science is fine for other things, but useless for audio.

There is no magic dividing line separating assessing audio qualities in either scenario! The assessment that a speaker is "bass-shy" when listening is no different in principle than assessing that a track I'm working on is "bass-shy" and needs an EQ boost. Or that the sound coming from a speaker is "emphasizing sibilance in an artificial manner" vs my assessing a production dialogue track has a treble boost causing precisely that same issue, and hence reaching for the EQ. It's the same principle.

And for that reason, if someone in the context of evaluating audio gear is arguing for the lack of validity and 'uselessness' of sighted listening or auditory memory, that claim is not magically walled off from any other real world stuff you'd have to explain. Like...if acoustic memory can not be relied on usefully, how do I manage to do my job using JUST that faculty? Etc. There may be liabilities in acoustic memory, but it is clearly not so unreliable as to be useless.

Again...none of that is to ignore the liabilities that research highlights regarding the variables that infest our perception. But we still have to place that in a wider context, to make sure we aren't overreaching with our skepticism to the point of incoherence.

 

[lowkeyoperations]

So we can safely conclude “listening sessions, in conjunction with detailed measurements, are a valid methodology for developing speakers”?

Hardly seems contentious to think you would listen to something that is specifically designed to be listened to. LOL.

 

[markus]

So we can safely conclude “listening sessions, in conjunction with detailed measurements, are a valid methodology for developing speakers”?

Well, that's how it's done currently. Massive room for improvement though. But as long as there's no adequate standardisation guided by additional psychoacoustic research and no audiophile asking for it, even arguing against it, it will remain the status quo.

 

[tuga]

What's the problem there?

Why don’t we rely on measurements instead of doctors and their albeit informed opinions?

 

[markus]

Why don’t we rely on measurements instead of doctors and their albeit informed opinions?

I just broke my wrist. My doctor (and I) were pretty happy we could rely on measurements in the form of a CT scan.

 

[lowkeyoperations]

Why don’t we rely on measurements instead of doctors and their albeit informed opinions?

Pretty sure we rely on both measurements and opinion.

You have a temperature. Probably a mild infection, not cancer.

 

[tuga]

Pretty sure we rely on both measurements and opinion.

You have a temperature. Probably a mild infection, not cancer.

It was a rhetorical question.

 

[dominikz]

That will depend on what is being asked of the memory.

Generally speaking: The more minute the audibility, the worse for acoustic memory. The more distinct the changes, the better.

I don't just rely on my immediate seconds-long short term memory (e.g. for changing volume or EQ on a sound). I also use my memory of sounds in my library, or that I have recorded, from yesterday, last week, last month, even years ago.

Just a few days ago I was creating a sound. I felt it needed a deep rumbling tone with a certain texture. I remembered creating just such a tone for a movie several years ago. I found it...it was just as I remembered and worked perfectly. People in my line of work do this all the time.

I'm constantly using my memory to recall sounds I've used before, or heard in my library, that will suit a scene I'm working on. For instance if I need a "dog barking" I have good recall of lots of the dog barks in my library, which allow me to find the one I want quickly. (One thing I often do before working on a movie is go through a relevant library and audition lots of the sounds. I can then often remember, often during months of work, a sound or recording I'd heard earlier that will suit my needs and go grab it).

If acoustic memory were truly "awful" or "bad," this wouldn't be possible.

This is why I keep emphasizing that you have to be cautious about what can be extrapolated, even from specific careful experiments. Those experiments don't show acoustic memory isn't useful.

IMHO we really need to make sure not to confuse the concepts of "signal" (in this context it is the original sound event) vs the concept of "noise" (any unwanted change of the signal, in this context linear and non-linear distortions introduced by the loudspeaker and room), to use the information theory terminology for a moment.

Let me clarify what I mean with this.

The human hearing apparatus and auditory perception are very competent at extracting the useful signal and ignoring the irrelevant noise, as well as remembering the main characteristics of the signal - this has obvious evolutionary advantages.
This is also why you can reliably recognize the voice of your friend over the phone even though it is transduced by a cheap microphone/loudspeaker and transferred via a band-passed, low bitrate codec though the telco network (all of which is introducing linear and non-linear distortions, and sometimes uncorrelated noise to the signal).
What you are recognizing is not primarily the 'frequency response' of the voice (there's really no such thing in this context), you are recognizing the unique sequence of spectral resonances / harmonics, specific correlated noise signature as well as signal envelope that make up the auditory signature of a specific human voice.
Of course other factors play here as well: e.g. rhythm, diction, phrasing and sets of language expressions characteristic to a person surely help as well. Similar mechanism allow you to distinguish a guitar from a piano, or to enjoy a favorite piece of music from a cheap BT speaker - i.e. sound source recognition in humans is relatively insensitive to frequency response deviations, non-linear distortion, uncorrelated noise and dispersion patterns.

It is also why you're able to recall the features of a sound event recording you heard a long time ago.

However when we're talking about loudspeakers (or microphones, amps, rooms, DACs...) we can really look at them as 'communication channels' rather than 'signal' - they are devices meant to relay (reproduce) the information / signal from a sound source (in this case the original sound event), but they can also introduce certain kinds of deviations to this signal (typically linear and non-linear distortions and uncorrelated noise).
As reproduction devices aiming for transparency these devices are not sound generating sources in a classical sense, with their characteristic set of high-level spectral resonances, specific envelopes and characteristic noise patterns that we're good at identifying (as is e.g. human voice, or a certain guitar, or a car engine), so the familiar human method of sound identification cannot really be applied.
When you are trying to identify loudspeakers (or most other audio equipment) you're instead trying to identify specific deviations that are common when reproducing any sound over it - i.e. the very thing that the human auditory system is struggling to tune-out to be able to focus on the all-important signal.

To return to the phone call analogy, trying to identify individual loudspeakers based on listening to various recordings is more like trying to identify the mobile phone model the person is using to make the call rather than the person using it! Correctly identifying the phone may of course be possible for devices that sound very different, but in many cases it may be quite difficult, especially between devices that perform similarly. However identifying the person on the other end is typically very easy.

Lastly, the main issue with sighted auditory comparisons is that they are unreliable, not that they cannot be accurate (of course they can). But how are we to know that the result of such comparison is relevant to us? Controlled (blind, level-matched) listening at least removes well-known sources of bias that can demonstrably skew such evaluations - this is why they are promoted. But even with basic controls in place, without statistical validity the results may still only really be relevant to the person doing the test. Someone's big difference may be irrelevant to others. In short - audio is evil.

Anyway, just my 2c (and sorry for the long-winded response )

 

[MattHooper]

Thanks for that reply dominikz!

As I've said, I think we have to keep the Big Picture in mind and not only identify why and when our auditory discretion or memory is unreliable, but also when it is reliable (especially for any practical purpose).

It seems your reply is an attempt to see that big picture. And I agree with a lot of what you wrote.

However, along the way you made a distinction that I don't think I agree with:

However when we're talking about loudspeakers (or microphones, amps, rooms, DACs...) we can really look at them as 'communication channels' rather than 'signal' - they are devices meant to relay (reproduce) the information / signal from a sound source (in this case the original sound event), but they can also introduce certain kinds of deviations to this signal (typically linear and non-linear distortions and uncorrelated noise).
As reproduction devices aiming for transparency these devices are not sound generating sources in a classical sense, with their characteristic set of high-level spectral resonances, specific envelopes and characteristic noise patterns that we're good at identifying (as is e.g. human voice, or a certain guitar, or a car engine), so the familiar human method of sound identification cannot really be applied.
When you are trying to identify loudspeakers (or most other audio equipment) you're instead trying to identify specific deviations that are common when reproducing any sound over it - i.e. the very thing that the human auditory system is struggling to tune-out to be able to focus on the all-important signal.

I don't really see this distinction. Sound is sound. Something either sounds different - identifiably so - or it doesn't. And it matters what we put our attention to in any instance.

If you take two people and ask them to speak the same sentence, they could be described as simply"reproducing the same information." But the will sound different when doing it. If the differences are gross enough, we'll notice how that sentence sounds coming from one speaker vs another.

Similarly if you had a recording of a person speaking that sentence, and play it through different speakers - it will sound different.

What is important isn't a distinction between live and reproduced or whatever. What is important in either case is that they sound different, and therefore the scale of the difference. Is it audible? Memorable? etc.

It's true that in everyday life we tune out "noise" to concentrate on sifting a signal. But that has to do with what we are concentrating on. For instance if we can have conversations in two different rooms, one quite "dry" of reverb, the other large and reverberant, we'd normally be concentrating on what someone is saying, our brain "filtering out" the acoustics in terms of our attention. But if we care about the acoustics, then we can pay attention and notice how different the sound of a voice is in each room.

In that respect, when audiophiles are listening to vocal tracks through different speakers, they are specifically paying attention to the differences in the presentation of those vocals through different speakers.

So I don't see an in principle divide. Sonic differences are there or not, memorable or not.

Another example from my job. I'm always including "room tones" for the backgrounds in my tracks for every indoor scene (and "air" for outdoor scenes). This helps the dialogue sit better in a track. Sometimes I'm matching the character of the room tone in the original production sound, often not. So for instance I've recorded a fairly/damped roomtone in my house, and I have manipulated it in various ways for various uses. We are talking very delicate subtle sound of the "air" moving in a room past the mic or any rumbles or buzzes it might pick up. Via EQ I've slightly rolled off the "air" or highs (and maybe taken out some bass) so as to create small spaces (like a closet) or increased bass (and/or added reverb) for larger spaces etc. So...it's the same track, a very subtle sound - not all the gross articulations or harmonics one would find in a voice - but mostly manipulated by EQ. And I can remember the sonic character of these different versions I created - at least well enough that I know when and where to use them.

So when it comes to speakers, while there is clearly all sorts of "noise" in terms of what can infect memory, can we apprehend, and remember some sonic impressions that have some level of "accuracy" over time? I think it's possible. Even if you might not be able to recall in your head the exact sound like a tape recorder, you may still have the generally accurate impressions - "really powerful bass" or "super clean and clear sounding" "free of apparent resonances" etc. (I often visit my friend who has two very different sounding systems, and they always sound different in exactly the way I remembered. It's my strong hunch I could identify which if comparing them in a blind test).

 

[dominikz]

As I've said, I think we have to keep the Big Picture in mind and not only identify why and when our auditory discretion or memory is unreliable, but also when it is reliable (especially for any practical purpose).

It seems your reply is an attempt to see that big picture. And I agree with a lot of what you wrote.

Thanks!

I don't really see this distinction. Sound is sound. Something either sounds different - identifiably so - or it doesn't. And it matters what we put our attention to in any instance.

So I guess the phone call analogy wasn't as compelling as I thought it was.
Joking aside, while sound is indeed sound, not all sound is processed the same way by the human auditory perception. It should e.g. be fairly obvious that we're much better at recognizing and remembering characteristics of voices than we are at recognizing sound signatures of different rooms. You could say that this is a result of conditioning/training and I might be inclined to agree - but note that this is 24/7 conditioning throughout your life. I.e. even if you are an audio professional you have much more 'training' tuning-out the unwanted noises and distortions that you have listening for them specifically

If you take two people and ask them to speak the same sentence, they could be described as simply"reproducing the same information." But the will sound different when doing it. If the differences are gross enough, we'll notice how that sentence sounds coming from one speaker vs another.

Please note that when I say 'signal' in my previous post I don't really mean the content of the message the human speaker is trying to deliver - I'm mainly talking about the sound of the voice. I.e. the spectral signature that makes a person's voice different than another ones.

Similarly if you had a recording of a person speaking that sentence, and play it through different speakers - it will sound different.

It will sound different, but in most cases you will still be able to recognize the voice. This is because the speakers added some 'noise' (information theory lingo again, in this example the 'noise' is mainly linear distortion and a change in sound dispersion characteristics, both due to the loudspeaker), but not enough to hide the 'signal' (the persons characteristic voice signature).

What is important isn't a distinction between live and reproduced or whatever. What is important in either case is that they sound different, and therefore the scale of the difference. Is it audible? Memorable? etc.

Sure - but my point is that humans are very good at identifying sounds that have consistently different spectrum and envelope characteristics (e.g. we're good at identifying the sound of a car engine vs the sound of the guitar string being plucked) - these differences are very memorable.
On the other hand, the same sound with two different linear distortions (e.g. small FR difference) overlaid is more difficult because 1) the characteristic spectral content is almost the same, and 2) this type of distortion the auditory system subconsciously filters-out all the time. As one example of this I suggest to lookup HRTFs and see how they change at different angles.

It's true that in everyday life we tune out "noise" to concentrate on sifting a signal. But that has to do with what we are concentrating on. For instance if we can have conversations in two different rooms, one quite "dry" of reverb, the other large and reverberant, we'd normally be concentrating on what someone is saying, our brain "filtering out" the acoustics in terms of our attention. But if we care about the acoustics, then we can pay attention and notice how different the sound of a voice is in each room.

Indeed. But I hope we can agree that it is still much easier to consistently and reliably recognize a voice than a room based on sound alone - even if you're a pro Voice recognition is fully intuitive and almost 100% correct - often even when the sound is quite distorted!

In that respect, when audiophiles are listening to vocal tracks through different speakers, they are specifically paying attention to the differences in the presentation of those vocals through different speakers.

Definitely, most of us on this forum are guilty of doing this. Some will of course be more successful that others, and training can help a lot with such tasks. But still it is much more difficult than recognizing the usual sounds we're surrounded with daily.

Another example from my job. I'm always including "room tones" for the backgrounds in my tracks for every indoor scene (and "air" for outdoor scenes). This helps the dialogue sit better in a track. Sometimes I'm matching the character of the room tone in the original production sound, often not. So for instance I've recorded a fairly/damped roomtone in my house, and I have manipulated it in various ways for various uses. We are talking very delicate subtle sound of the "air" moving in a room past the mic or any rumbles or buzzes it might pick up. Via EQ I've slightly rolled off the "air" or highs (and maybe taken out some bass) so as to create small spaces (like a closet) or increased bass (and/or added reverb) for larger spaces etc. So...it's the same track, a very subtle sound - not all the gross articulations or harmonics one would find in a voice - but mostly manipulated by EQ. And I can remember the sonic character of these different versions I created - at least well enough that I know when and where to use them.

I have no doubt this is true - we all make a lot of intuitive connections in our daily jobs that make our workflow more efficient. Also, noise-like sounds can of course be 'signals' as well

So when it comes to speakers, while there is clearly all sorts of "noise" in terms of what can infect memory, can we apprehend, and remember some sonic impressions that have some level of "accuracy" over time? I think it's possible. Even if you might not be able to recall in your head the exact sound like a tape recorder, you may still have the generally accurate impressions - "really powerful bass" or "super clean and clear sounding" "free of apparent resonances" etc. (I often visit my friend who has two very different sounding systems, and they always sound different in exactly the way I remembered. It's my strong hunch I could identify which if comparing them in a blind test).

It may be possible that people could be trained and may be able to remember such information - but I'm sceptical. And I definitely wouldn't expect most people to be successful at this - even if they are audiophiles or audio professionals. IMO this doesn't really take away from their competence - it is just a difficult task that is surely very error-prone.

EDIT: Of course identifying two sufficiently different loudspeakers in an double-blind ABX with instant switching should be quite easy. But on the other hand successfully identifying a loudspeaker in a blind test based on memory and without a reference is IMHO unlikely.

 

[fluid]

Caution against removing sources of bias in an experiment is always good advice, but if taken too far it can easily give the impression that without the strictest controls a test is useless. If the worst of them can be removed then basic subjective listening test become useful, particularly when designing a product to be listed to.

As an example of both bias and auditory memory, if I call my mum at a time she is not expecting me to, often it takes here a minute to work out who is calling. The expectation bias overrides the auditory memory.

When setting up my speakers I listened to different filters based on different measurements or EQ/window settings to try and work out which one I like more. I have seen the measurements before hand, I know something about the things I am testing that could impart bias. I try my best to make sure I don't know which of the two I am listening to and switch backwards and forwards as many times, for as long as I need to. Once I'm satisfied there is a winner I look to see which one it was, sometimes I'm surprised as the one I thought I would like more loses out. I wouldn't try and extrapolate this data as some kind of objective proof, but it's enough for me to to be happy that I found what I actually think sounds better at that time in that place.

I would love there to be more psychoacoustic testing to help guide design in the right direction but as time goes on I am more and more convinced that there is no universal single right answer for everybody. Preference in music content and imaging added to experience of listening in specific environments create some really difficult problems for a designer to overcome. Some will pick the best overall compromise accepting that it is not perfect in any one way but satisfactory in all, other's will sacrifice one thing for another. Sometimes people shift preference depending on the scenario.