Can useful knowledge be gained via subjectivity?

[Doodski]

Why is it that whenever someone puts something new into a system, the results are always that "the soundstage opened up and disappeared to beyond the speakers... It was like a blanket was lifted"? Why is that always what happens? LOL. It's called expectation bias frankly...

I just got to the end of Paul McGowan's book "99% True" for the second time. And when he starts blathering on about subjectivity and not being able to measure certain things I think he is truly peddling in snake oil. In fact he pretty much said exactly what I put in those quotes in the previous paragraph. Go figure...

How was the book in it's entirety and in general?

 

[Newman]

Well if it includes the Dance of The Seven Veils....

 

[solderdude]

IF an equipment has symmetrical layout for both L/R channels, differences to high loose tolerance is going to have an effect as well.

Yes, very measurable for sure.
3% difference is detectable under blind level matched conditions. Don't use 10% or 20% tolerance resistors (if you can find them)
You should realize that designs use lots of resistors and this means tolerances average out but you can have bad luck and it goes in one way or the other.
Most electronic circuits use overall feedback this means that, aside from the 2 resistors used in the feedback path the rest of them is compensated for.
You should also know that when I grab a box of 5% resistors most of them will be below 2% or even 1%.
For instrumentation devices you can buy 0.01% tolerance resistors. These are often simply resistors from a 0.1% production line that measured below 0.01%.

Most L-R differences come from the volume control (a mechanical one)
This can skew L-R balance but won't affect sound quality.

For speakers (eg. tweeter resistor), changes in the resistance esp. due to temperature will affect the signal amplitude. This is not "resistor sound".

All resistors have a temp coefficient. It can be positive or negative and manufacturers can make this low playing with materials.
You can design an XO with resistors with too small power ratings to save a $ or 2 or save space. There are manufacturers that do.
For this change to reach audible levels you need quite a temperature change so substantially under rated resistors.

Btw, resistors have 'noise' as well

Yes, all electronic components do.

You noticed the R in there ?

And yes, using a 1M or 100k Carbon composite resistor at the input of a mic/MM/MC input is not the wisest thing to do when you want a decent S/N ratio.
A low ohm metal film (which costs about the same if not cheaper) would be a better choice.

Do you believe that when you replace the resistors used in a somewhat noisy amp with Vishay foil resistors the performance will (audible) increase or will measured results not change much as the noise in there is mostly determined by the semiconductors and the values on the input and feedback resistors ?

 

[MattHooper]

Why is it that whenever someone puts something new into a system, the results are always that "the soundstage opened up and disappeared to beyond the speakers... It was like a blanket was lifted"? Why is that always what happens? LOL. It's called expectation bias frankly...

I just got to the end of Paul McGowan's book "99% True" for the second time. And when he starts blathering on about subjectivity and not being able to measure certain things I think he is truly peddling in snake oil. In fact he pretty much said exactly what I put in those quotes in the previous paragraph. Go figure...

Actually I have had a bit of fun trying different things recently, different tubes in my amps - if anything I often perceived worse sound - and since I had them around from another project I put some isoacoustic footers under my speakers, and I didn't perceive the sound to lift veils or open up, if anything the opposite (so I took them out). In another instance I had tried some cheap spring footers under my speakers to decouple them from the floor that seemed to indeed "open up" the sound, make the speakers "disappear more" and all that. Except the sound also seemed to thin out too much (which could have been just my imagination, or it could have been simply from raising the speakers a bit higher on those footers, or some combination). I this led me to do a trial of some Townshend speaker bars, spring based but more purpose-built for decoupling speakers and they do not raise the height of the speakers. What I hoped was that I'd get some of the "speakers disappearing/bass tightening" etc I seemed to get from the other spring footers, but with less change in tonal balance (speakers not changing height). But I didn't end up getting the effect I hoped for and sent them back.

Now, every bit of that could be some form of bias or imagination effect at work. But my point in those examples is simply that the "expectation effect" - "you hear it because you expect to hear it" doesn't explain all of them. (Again, which doesn't mean it's not perceptual error). But "Placebo Effect" and "Expectation Effect" is so often raised by skeptics that the purely subjective audiophiles seize on to this to just bolster their beliefs. Every time they have an experience that they "didn't expect" they take it as verifying the reality of their perception. As in "Actually, I tried X and DIDN'T expect to hear a difference, but I did!" or "I tried some cables and expected the most expensive to sound best but they didn't, the cheaper ones sounded better, so it wasn't my imagination Mr. Skeptic!"

But of course our biases and perceptual errors can come in so many other forms that a strict "expectation effect."
For instance, the mere act of listening for a difference, even if you don't expect one, can change your perception making it seem you are hearing a difference. That's why ANY of us can fall for bias effects. You can have "unexpected" experiences with bias.

I find I've had to point this out ad infinitum in the subjective forums because The Placebo Effect and Expectation Bias seem to be what anyone thinks of as variables, when it's much more varied and complex.

 

[Robin L]

Why not deconvolve it?

Had to look up that term.
I don't know if you've done any recording. Recording engineers/producers pick a given set of microphones because they think the coloration of the microphone fits the music they're recording. While there are some folks motivated to explore different ways of recording, those are aren't the folks making bank as recording engineers or producers. As far as I can tell, the concept makes sense for forensics, but recording engineers don't have an incentive to change working practices. I don't see any meaningful changes until such time as the physical design of microphones changes, until such time as the tone color and other aspects of the microphones themselves changes. I can imagine some device that measures sound without a tympanic membrane and its associated resonances might appear. Besides which, most audiophiles are happy enough with the results achieved by more conventional means. Another thing, I would expect that deconvolution would become much harder as the number of microphones increases. Purist 2-microphone stereo is not the technique used for the vast bulk of commercial recordings. Even classical music recordings of relatively simple music usually uses more than one set of microphones, frequently mixing microphone types in the process.

 

[escksu]

Yes, very measurable for sure.
3% difference is detectable under blind level matched conditions. Don't use 10% or 20% tolerance resistors (if you can find them)
You should realize that designs use lots of resistors and this means tolerances average out but you can have bad luck and it goes in one way or the other.
Most electronic circuits use overall feedback this means that, aside from the 2 resistors used in the feedback path the rest of them is compensated for.
You should also know that when I grab a box of 5% resistors most of them will be below 2% or even 1%.
For instrumentation devices you can buy 0.01% tolerance resistors. These are often simply resistors from a 0.1% production line that measured below 0.01%.

Most L-R differences come from the volume control (a mechanical one)
This can skew L-R balance but won't affect sound quality.



All resistors have a temp coefficient. It can be positive or negative and manufacturers can make this low playing with materials.
You can design an XO with resistors with too small power ratings to save a $ or 2 or save space. There are manufacturers that do.
For this change to reach audible levels you need quite a temperature change so substantially under rated resistors.



Yes, all electronic components do.

You noticed the R in there ?

And yes, using a 1M or 100k Carbon composite resistor at the input of a mic/MM/MC input is not the wisest thing to do when you want a decent S/N ratio.
A low ohm metal film (which costs about the same if not cheaper) would be a better choice.

Do you believe that when you replace the resistors used in a somewhat noisy amp with Vishay foil resistors the performance will (audible) increase or will measured results not change much as the noise in there is mostly determined by the semiconductors and the values on the input and feedback resistors ?

Cheers!! I wish this forum has more pple like you instead of those who simply shoot down others and claim that they are the same and it doesn't make a difference.

REgarding your question, I have never tried using vishay foil resistors so I can't answer it. However, I do personally believe its not going to make a big difference. However, its still fun to try it out and learn something new. Although 1 single change is not likely to have any meaingful difference, sum of these changes could add up.

 

[solderdude]

When folks state changing resistors for other types or using foil instead of ceramic and claim the sound quality improves is what is questioned.

I think most ASR members are fully aware that all cables, resistors, capacitors, transistors, IC's and power supplies all have different properties which can be shown by measurements quite easily.
The point they might want to bring across is that while measurements may show some changes in measurable limits certain changes are inconsequential as they do not reach any audible levels.

The difficulty lies in the fact that to really prove something tests have to be done in certain scientifically sound ways instead of changing something and then stating... well this or that component change gave (clearly) audible changes in this or that area.

Asking to test differently (more scientifically correct) or having asked this many many times can make people react in an, annoying to the one making a claim, fashion and being put of by this. They expect (and sometimes demand) to be taken seriously.
Usually these individuals do not want to or believe in testing in any other way which doesn't help.

May be seen as 'toxic' to people who believe everything matters and hearing can detect so much more than measuring. Mostly because they do not fully understand measurement results and methods.

In the end the vast majority of music listeners don't give a crap about measurements but know what they like or don't.
They should look and discuss in other fora. When they want to learn something and maybe get a different perspective they should stick around, read, ask questions and debate (in a friendly way) and take some advice.
In the end.. they will be free to believe whatever they want but at least get to see things from more than just the 'subjective side'.

 

[pkane]

OK, regarding the music, what I am saying is that it cannot be measured objectively (favourite word in the forum). Eg. THD, SINAD etc... they are measured using equipment and it return certain values. We can't measure music that way. Maybe the better word to use is genre. Whether the person enjoy listening to the music/song is entirely up to the person's own experience and preferences. This is where its entirely subjective.

Music can be measured, although THD/SINAD, etc. are not the correct values for this -- these are for measuring electronics. If you mean "music enjoyment" can't be measured, then that's also incorrect. Many studies measure the subjective enjoyment of music, drinks, food, automobiles, movies, etc.

 

[solderdude]

I think @escksu means that when you play music A, B and C there is no electronic analysis that will show (without a shadow of doubt) that persons X, Y and Z will like it or not and what they like about the analyzed song.

Of course, google might know what kind of music you like and through its database will probably be able to tell you which music person X, Y and Z will like and make some excellent suggestions.

 

[pkane]

I think @escksu means that when you play music A, B and C there is no electronic analysis that will show (without a shadow of doubt) that persons X, Y and Z will like it or not and what they like about the analyzed song.

Of course, google might know what kind of music you like and through its database will probably be able to tell you which music person X, Y and Z will like and make some excellent suggestions.

Of course not, why would anyone think that this type of precision is possible? Individual tastes are based on genetics, upbringing, culture, experience, environmental factors and many other things that are very different for each person. That said, I remember reading a study that linked some of the music preferences to the language the mother spoke while the baby was still in the womb

 

[antcollinet]

In the interest of good faith discussions...and getting all geeky/philosophical...

First, as it's often been acknowledged, pretty much no one here thinks subjectivity is irrelevant. The outlook of a site like this is trying to correlate measurements to subjective perception in a reliable fashion. So with that out of the way, I'm presuming the real meat of the question is whether *purely* subjective experience - that is without appeal to measurements, and perhaps without even controls for bias - can be a source of useful knowledge.

Can useful knowledge be gained via subjectivity?​

Broadly speaking: of course. Essentially all our empirical inferences come to us through our subjective experience and perception. When I went to leave my house yesterday I was stopped by a some netting and signs from walking down my steps. The steps were gone. Our contractor, building us new steps, took them away and put up a warning. I gained knowledge of this via my subjective perception, and if I simply ignored it I could have broken a bone stepping off the porch with no steps.

That may seem like an obvious thing to point out about the use of our subjective perception, but it's actually something that I find starts to get obscured sometimes in conversations even about audio gear when purely subjective impressions and reports are dismissed as wholly useless.

I think there's lots of room for nuance and "it depends" and precisely what a person may define as their goal.

For audio as in anything else: There certainly is a sense in which purely subjective impressions are "useless." IF what you desire is a scientific level of confidence, where correlating measurements, controlling for variables like bias etc can result in justifiable confidence levels in a conclusion (e.g. what may be audible or not, or what may be predictably preferred by a percentage of listeners etc) then anything offered that lacks in that method is indeed useless. What some audiophile says he "heard" just isn't going to meet that bill. And that is of course, in many instances, a totally reasonable goal or criteria to have.

But having that criteria doesn't automatically entail that we can't have some sense where exchanging purely subjective accounts is, to some degree, reasonable and useful. After all, every day we navigate the world mostly successfully by intersubjective engagement. I reported my perception of the stairs being gone from our porch to my wife, so she wouldn't try going out the front door. Did we need to do a study on the phenomenon of our porch, and the reliability of our perception, in order for it to be reasonable to exchange information this way? That would be untenable.

So while it's true that purely subjective reports are subject to error and hence a "scientific" approach is going to be obviously more reliable, it's also the case that our perception seems reliable enough to get us through the day, and in exchanging lots of intersubjective information, and so when we want to totally dismiss subjective perception as "useless" we should be ready to examine just how far we want to take it, and why.

For instance: should people refrain from discussing things like the sound of different mastering jobs on various albums, simply from having listened to them? Or even the different production method and sound of different recordings? If so...where does this almost Cartesian skepticism of purely subjective impressions stop and why, and how do you make life tenable?

If we rejected every subjective perception that wasn't scientifically vetted, life would be impossible.

But if it IS reasonable under many circumstances to exchange subjective impressions as relatively informative, e.g. discussions of different record masters or recording techniques etc, WITHOUT requiring a truly rigorous scientific account for our every impression - what could justify this?

I've argued that the simple heuristic that we all naturally carry around with us: "Extraordinary claims require extraordinary evidence," helps us navigate this space. (Thanks Carl Sagan!)

If someone where I live tells me he saw a racoon in his backyard last night, it may not be a scientific account - and he could have been wrong, seeing things, been mistaken etc - but it remains reasonable for me to accept the account, provisionally, because of it's plausibility. If he had reported his perception that he saw Big Foot or a T-Rex stomping around the yard, then it's more reasonable to say "Gonna need some more rigorous evidence for that."

I take the same approach to discussions of the sound of albums, mastering, speakers etc. If someone tells me he came from a speaker audition and found the speaker to sound very thin in the midrange, or boomy and unbalanced in the bass etc, I can accept his report, provisionally, given that it's entirely plausible that a speaker could sound like that. Whereas if he reported the same thing for exchanging an audiophile Ethernet cable for a regular functioning Ethernet cable, that goes in to "extraordinary claim" in terms of it's technical plausibility, and I'll want more rigorous evidence.

It's true that even when there are audible differences (e.g. speakers), just like when there are objective differences in what we are seeing, our perception can still be affected by bias. Error is always hovering over our perceptions. But that doesn't mean it isn't reasonable to provisionally accept subjective reports, although with an accompanying lowered confidence level.

So using subjective impressions of speakers as an example: If I see a certain audiophile on a forum, or a certain audio reviewer, seems to be noting some of the things I really care about in reproduced sound, that gets my attention. After all, speakers quite plausibly sound different. And especially if I have found this person's reports on other speakers I've heard seem to coincide with my own impressions, that further suggests we are impressed with the same things. I have been led to some very happy speaker purchases via the combined reports of some audiophiles and reviewers, when I saw they seemed to care about what I cared about, and converged identifying it in a product. Where I have found the product to have just those characteristics reported by others. So the purely subjective reports of other listeners have indeed been "useful" for me. The reverse has been true, where others have said they found my descriptions of speakers very useful - they heard what I described once they had a chance to hear the same speakers.

(And as I've pointed out before, I've also found that other people's, or a reviewer's, subjective reports matched my own impressions even when I heard a speaker first, and read the reports afterward).

So my view is that to some degree, the usefulness of swapping purely subjective reports is going to depend to some degree on the person.
In some cases, something is only "useful" if you care to use it.

If someone is entirely dismissive of subjective descriptions of sound, and instead educates himself on speaker measurements to the point he can predict what he'll like from measurements and only wants the reliability that comes from listening tests controlling for bias..then that person will rightly have no use for purely subjective reports.

But that person is also the least likely to find use in subjective descriptions, even if there is any use (as we see many audiophile descriptors are waved away as mere fantasy).

So having the goal of the most reliable method of inference doesn't mean intersubjective "knowledge," exchanging purely subjective perceptions, is wholly without worth or useless.

Again, the more skepticism you cast on subjective perception - in audio or anywhere else! - the more you take on the burden of answering when and why it's ever a reasonable way to exchange information subjectively. (And I've yet to see how you get out of this slippery slope of skepticism without appealing to essentially the heuristic "extraordinary claims require extraordinary evidence" and it's obverse "ordinary claims only require ordinary evidence" I have mentioned).

If I had the energy - that is exactly what I'd have written. Top Post.

 

[krabapple]

OK, regarding the music, what I am saying is that it cannot be measured objectively (favourite word in the forum). Eg. THD, SINAD etc... they are measured using equipment and it return certain values. We can't measure music that way. Maybe the better word to use is genre. Whether the person enjoy listening to the music/song is entirely up to the person's own experience and preferences. This is where its entirely subjective.

Let's posit that between listening session X and Y nothing about the music has changed. If you claim to be enjoying the music more in session B than session A, the place to look for reasons isn't really the music. It's in the system and in your head. If it's in the system it can be measured. If it's in your head and not in the system...who cares?

I surely don't, and on an audio science forum shouldn't be expected to.

 

[Holmz]

Let's posit that between listening session X and Y nothing about the music has changed. If you claim to be enjoying the music more in session B than session A, the place to look for reasons isn't really the music. It's in the system and in your head. If it's in the system it can be measured. If it's in your head and not in the system...who cares?

I surely don't, and on an audio science forum shouldn't be expected to.

But the forum membership may be expected to care.
If we pretend that psychoaccoustics are not a thing, then that does not comport with reality.

As science has some ties with comporting to reality, it seems that even psychology is within the realm of science and audio.

So we care in as much as is needed, in order to weed through the morass.

 

[Raindog123]

I’ve stumbled over this post earlier today. @JohnBooty ‘s observation about the very origin of the Harman Curve (that we use as a bonafide objective reference all the time!) - being the averaged subjective preference - might be a good answer to this-thread question.

 

[krabapple]

But the forum membership may be expected to care.
If we pretend that psychoaccoustics are not a thing, then that does not comport with reality.

As science has some ties with comporting to reality, it seems that even psychology is within the realm of science and audio.

So we care in as much as is needed, in order to weed through the morass.

As my posting history indicates, I very much think psychoacoustics is a thing.

Psychoacoustics is one of the sciences that tells us how mistaken our beliefs about audio can be, when they are based on subjective impressions with no bias controls in place.

 

[krabapple]

I’ve stumbled over this post earlier today. @JohnBooty ‘s observation about the very origin of the Harman Curve (that we use as a bonafide objective reference all the time!) - being the averaged subjective preference - might be a good answer to this-thread question.

All studies of preference -- indeed all investigation of audible difference , too -- depend on reports of subjective experience. This isn't a revelation.

The important factor that makes it scientific research, is how well controls are put in place so that the reported preference/difference is derived from what's audible, and not something else.

You're mistaken to think any curve 'objectively' sounds the best, or that anyone with a clue actually makes that claim. What 'we' do is acknowledge good research showing that listeners prefer some curves over others on average. And we acknowledge the good bet that we, too, are 'average'.

 

[JRS]

Why not deconvolve it?

Oh, but to do so would require digital audio.

 

[coonmanx]

How was the book in it's entirety and in general?

Great book. I am almost done reading it for the second time now.

Lots of great history in there as well. Like when he almost got to record Pink Floyd live.

 

[Newman]

Oh, but to do so would require digital audio.

 

[richard12511]

People in ASR ONLY cares about measurements..... ITs a KNOWN FACT.

Not true at all. So no, not a known fact at all. In my view measurements are only useful insofar as they can predict subjective preference. Subjective preference is ultimately the most important factor for me. Measurements just happen to be far better at predicting it than other people's sighted subjective impressions are.

ASR is probably the least "toxic" forum I've ever been a part of. They're far less hostile than any other forum I've been on when it comes to ideas that go against the forum's beliefs. I think maybe you just see those other (more subjective) places as less toxic simply because their views more closely align with your own.

Other forums like Audiogon or SBAF are far more hostile towards talks of "measurements" than ASR is against subjective impressions.