Resolve's B&K 5128 Headphone Target - you can try the EQ's.....

[Deleted member 16543]

No. I call this ad-hoc or DIY/hobby level testing.

So.. if somebody else does exactly what you did, but with a rig that looks different from yours (which looks similar but measures just as different from Harman's rig as a 5128, in terms of deviation), your results are still more valid than theirs?
Why?

 

[Deleted member 16543]

Do you mean constant vs. frequency, or...? Like, we know the input Z of human ears - that's what Jønsson et al, and long, loooong ago Rasmussen et al were about. I may be misunderstanding you here?

I meant if we could model the canal as a constant lump parameter that doesn't change with incoming sound direction or source distribution in space. It would still have its own frequency response, of course, which has a reactive component, but always the same for every possible measurement.
I don't think we can demonstrate this is the case, though, so a fully anatomically accurate geometry is, at the very least, a safer bet.
Again, my stance is that it is actually necessary.

 

[Mad_Economist]

which looks similar but measures just as different from Harman's rig as a 5128, in terms of deviation

At high frequency, to be clear - at low frequencies, because the error is not about ear Z or pinna geometry but rather leakage response, both of the 45CA datasets Sean has (the KB501x and the modded Welti pinna used in Harman's work) are quite similar. Of course, the 5128 actually comes closer to the humans in Sean's dataset, and it's very much possible to notice and correct incorrect leakage during measurements, so...

I meant if we could model the canal as a constant lump parameter that doesn't change with incoming sound direction or source distribution in space. It would still have its own frequency response, of course, which has a reactive component, but always the same for every possible measurement.
I don't think we can demonstrate this is the case, though, so a fully anatomically accurate geometry is, at the very least, a safer bet.
Again, my stance is that it is actually necessary.

The frequency response of the ear is, because the ear's impedance is reactive, variable based on source impedance. Like, it's better to view the headphone-ear system as having a single frequency response (measured at one point), which is itself the product of the interaction of their impedances. Because of this, the "ear FR" (or, at least, the delta of the FR you'd measure with and without the ear load) will vary any time the acoustic source impedance varies.

 

[amirm]

If I'm to understand you correctly, does this mean you consider your own results to be only DIY/hobby level?

You asked my opinion of this work and I gave it. Now you are arguing with me about it? If you didn't want or value my opinion, you should not have asked.

Regardless, if I asked Sean if my work is built on top of this research and is very compatible with it, I am pretty sure he would say yes. Whereas if I asked about what you are doing, he will likely be quite reserved.

It seems that you have so trivialized this effort as to want anything you do to be valid and authoritative. Well, it isn't that way. It just isn't in my book as evidenced by what started this thread. You do understand all the problems with this type of survey, yes?

 

[Mad_Economist]

You asked my opinion of this work and I gave it. Now you are arguing with me about it? If you didn't want or value my opinion, you should not have asked.

I'm not being sarcastic nor attempting to argue here, I'm trying to understand the burden of evidence you're requiring for disparate measurement systems.

Regardless, if I asked Sean if my work is built on top of this research and is very compatible with it, I am pretty sure he would say yes. Whereas if I asked about what you are doing, he will likely be quite reserved.

As it happens, I'm setting up a dialogue with Sean at present - would you like to be included?

It seems that you have so trivialized this effort as to want anything you do to be valid and authoritative. Well, it isn't that way.

This seems bit impolitely framed, honestly. I - and, to the best of my knowledge, @Resolve and everyone else at headphones.com - am not setting out to trivialize anything, nor to demand that our efforts be treated as "authoritative" (whatever that really means in this context) - I want to make sure our reasoning is understood, and to understand what other people are saying in response.

It just isn't in my book as evidenced by what started this thread. You do understand all the problems with this type of survey, yes?

Certainly, I'm paying...virtually no attention to that thread (sorry @Robbo99999!), but it's just one very small piece in the project. We wouldn't implement changes without a better controlled test, and we aren't proposing that this thread represents some grand takedown of Sean's work. @Resolve has a bee in his bonnet that, after Lorho 2009, the height of the ear gain peak should be adjustable separately from other "slope" parameters. I'm not entirely sold on this reasoning, and it's something we'll be working on more proper tests for, but I believe his intention in creating that thread was quite literally just to confirm that it wasn't just him alone who felt that a small notch filter produced a positive effect.

 

[amirm]

As it happens, I'm setting up a dialogue with Sean at present - would you like to be included?

No. No interest. Even in the best case scenario I see this effort as a negative one resulting in confusion in the market. That headphones.com plans to put parallel measurements out for the same headphone points to this very worry. Good luck.

 

[Mad_Economist]

No. No interest. Even in the best case scenario I see this effort as a negative one resulting in confusion in the market. That headphones.com plans to put parallel measurements out for the same headphone points to this very worry. Good luck.

Alright! We'll keep working on it, and I'm hopeful that you'll be positively surprised with what comes out of it!

 

[Robbo99999]

uh... why does Mad Economist no longer have the Technical Expert badge?

Ah!

I'm not being sarcastic nor attempting to argue here, I'm trying to understand the burden of evidence you're requiring for disparate measurement systems.

As it happens, I'm setting up a dialogue with Sean at present - would you like to be included?

This seems bit impolitely framed, honestly. I - and, to the best of my knowledge, @Resolve and everyone else at headphones.com - am not setting out to trivialize anything, nor to demand that our efforts be treated as "authoritative" (whatever that really means in this context) - I want to make sure our reasoning is understood, and to understand what other people are saying in response.

Certainly, I'm paying...virtually no attention to that thread (sorry @Robbo99999!), but it's just one very small piece in the project. We wouldn't implement changes without a better controlled test, and we aren't proposing that this thread represents some grand takedown of Sean's work. @Resolve has a bee in his bonnet that, after Lorho 2009, the height of the ear gain peak should be adjustable separately from other "slope" parameters. I'm not entirely sold on this reasoning, and it's something we'll be working on more proper tests for, but I believe his intention in creating that thread was quite literally just to confirm that it wasn't just him alone who felt that a small notch filter produced a positive effect.

(Hey, that's alright, I just started this thread to link to Resolve's thread over on headphones.com & make people aware of the work being done over there, and this forum is a place where that project can be discussed by people here. "That thread" is the one over on headphones.com anyway, but that might be what you mean too.)

 

[Mad_Economist]

(Hey, that's alright, I just started this thread to link to Resolve's thread over on headphones.com & make people aware of the work being done over there, and this forum is a place where that project can be discussed by people here. "That thread" is the one over on headphones.com anyway, but that might be what you mean too.)

Sorry, to be clear, I meant Andrew's thread on headphones.com - as you can see by my posting, I've been fairly active here.

From my POV, the thread on the forum serves as a little data point about preference, which points to the merits of at least testing @Resolve's pitch to let users adjust ear gain separately from the bass and treble. I'm not sure this testing will produce anything interesting - but I would have said much the same about Miller & Downey 2022's inquiry into a high-frequency peak filter, and so I'd have been wrong there!

(attached is the publicly available version Knowles put out)

 

[Deleted member 16543]

The frequency response of the ear is, because the ear's impedance is reactive, variable based on source impedance. Like, it's better to view the headphone-ear system as having a single frequency response (measured at one point), which is itself the product of the interaction of their impedances. Because of this, the "ear FR" (or, at least, the delta of the FR you'd measure with and without the ear load) will vary any time the acoustic source impedance varies.

Ok. Let's start from scratch and see if we can get on the same page.
You are modeling a headphone measurement as an electrical circuit (that's ok, I'm on board). The circuit shows V1 as the test signal, Zsource as a model for the headphone impedance (dependent on positional variations too, I guess, otherwise we would also need an additional load after that?), Zload as the impedance of the ear, and Vout as the pressure sensed by the eardrum (mic capsule), correct?
What I'm asking you is if you think that Zload can be split into two components (R-L-C subnetworks), one of which (Zc) can be attributed solely to the eardrum and the other (Zp) to the pinna and geometry surrounding it.
And also, more to the point, if you think that the parameters of Zc (the values of its R-L-C network) remain constant with different measurements.
I believe you think they do, although personally it's an assumption I'm not agreeing with.
This is basically were we interpret the variations shown in the graphs of the research you linked to in different ways (you as noise, me as indication of something else happening).

I think the load of the canal varies with each measurement. In other words, different incoming sounds engage the canal in ways that require a different canal load model every time, to model the sensed pressure. In fact, I think that Zload can't be split into two separate components at all, especially at higher frequencies.
I think the more sensitive approach is to to think of the network to model Zload as having R-L-C values too entangled to separate them neatly into the two subloads (because of the complexity of the interaction between pinna and canal as incoming sound direction changes).

But IF that separation could be done, and IF Zc's R-L-C components had constant values across all measurements, then one could use a coupler with the same impedance Zc which doesn't necessarily have to maintain the canal geometry to perform equivalent measurements (but then again, as you already pointed out, IEM measurements would be somewhat wacky).

Is this a correct summary of each other's positions?

 

[solderdude]

I think the oversimplified schematic drawn was not meant as an exact description but to make a general point and used in order to make a simple plot as visual aid.
The actual acoustic load (impedance) will be much more complex than a simple LCR.
Not related to the headphone impedance.

 

[Deleted member 16543]

You asked my opinion of this work and I gave it. Now you are arguing with me about it? If you didn't want or value my opinion, you should not have asked.

Regardless, if I asked Sean if my work is built on top of this research and is very compatible with it, I am pretty sure he would say yes. Whereas if I asked about what you are doing, he will likely be quite reserved.

Your rig being different from the one used with Harman's research, it requires an adjustment of the target curve (different rigs require different targets).

So why is investigation into a target fit for the 5128 not the same as your investigation into how the target for your rig has to be adjusted from the optimal target of Harman's research (which incidentally keeps changing quite often itself)?

And why is your way of doing things and checking for your method's validity according to feedback from people that try your correction EQs that much superior to what Headphones.com (and any other 5128 employing team) could EVER muster to come up with?
We still haven't seen but a fraction of what they're working on behind the scenes, yet you seem so eager to shut down every 5128 based attempt at bettering that measurement-preference correlation.

Can't you give people the benefit of the doubt and wait for their final results first?
These are quite smart people we're talking about.
Be patient.. wait to see what they come up with first. Then you can trash them all you want.

Nobody is arguing you're not providing the best possible value for your readers at this time.
Just that there are promising indications that this may change in the future.

 

[Deleted member 16543]

I think the oversimplified schematic drawn was not meant as an exact description but to make a general point and used in order to make a simple plot as visual aid.
The actual acoustic load (impedance) will be much more complex than a simple LCR.
Not related to the headphone impedance.

Yes. Every load has multiple R-L-C components. That's why I called them networks.
What they actually look like I have no idea. I just know that being passive loads they can be modeled with those 3 basic components (a bucketful of each one of them per load, probably)

 

[solderdude]

That would look nice in a scientific article with tons of impressive equations but I see very little practical use for that acoustical simulation wrt real world headphone measurements with all their 'practical problems'.
Time can be spent better IMHO.

 

[markanini]

@Mad_Economist I'm sure the Reactive Abuse is not intentional on your part. Usually, it's something that sneaks its way into someone's behavior if they've ever had someone in their life who acted like that, and during moments when you're passionate about something. Everyone benefits from you being the canon of knowledge that you are. Consider, for a moment, that your excitement for new technology might be getting the best of you and causing the conversation with @amirm to take some unnecessary turns. Armin is giving reasons based on experiences with past efforts, and analysis, which needs the full acknowledgment it can get before even discussing tone or good will. The latter will stop any productive discussion from happening, don't be surprised by this.

Let's all make sure that we're not above empathizing with the guy that invested $50000 in a rig that turned out not to be viable.

 

[Deleted member 16543]

That would look nice in a scientific article with tons of impressive equations but I see very little practical use for that acoustical simulation wrt real world headphone measurements with all their 'practical problems'.
Time can be spent better IMHO.

Just to be clear, I'm not trying to find the correct network to model headphones measurements.
Mine is just an inquiry on whether Mad_Economist thinks the canal part of that load network is independent from its pinna and head part, and if its components (however many there are and whatever that subnetwork looks like) are constant across different measurements, in his opinion.
Like you, I have better things to do than to actually model that network in its details. I wouldn't even know where to begin, to be honest.

 

[isostasy]

@Mad_Economist I'm sure the Reactive Abuse is not intentional on your part. Usually, it's something that sneaks its way into someone's behavior if they've ever had someone in their life who acted like that, and during moments when you're passionate about something. Everyone benefits from you being the canon of knowledge that you are. Consider, for a moment, that your excitement for new technology might be getting the best of you and causing the conversation between you and @amirm to take some unnecessary turns. Armin is giving reasons based on experiences with past efforts, and analysis, which needs the full acknowledgment it can get before even discussing tone or good will. The latter will stop any productive discussion from happening, don't be surprised by this.

Let's all make sure that we're not above empathizing with the guy that invested $50000 in a rig that turned out not to be viable.

I've been following this thread quietly but have to speak out against remotely analyzing the psychology behind somebody's posts. Terms like 'Reactive Abuse' and attempts to read into past interpersonal relationships as impacting somebody's discourse on an audio forum is not appropriate.

I've found everyone's contributions interesting and some have taken considerable effort providing evidence and explanations. Disagreement exists but there's absolutely no need to read further into it than that.

 

[Chocomel]

@Mad_Economist I'm sure the Reactive Abuse is not intentional on your part. Usually, it's something that sneaks its way into someone's behavior if they've ever had someone in their life who acted like that, and during moments when you're passionate about something. Everyone benefits from you being the canon of knowledge that you are. Consider, for a moment, that your excitement for new technology might be getting the best of you and causing the conversation with @amirm to take some unnecessary turns. Armin is giving reasons based on experiences with past efforts, and analysis, which needs the full acknowledgment it can get before even discussing tone or good will. The latter will stop any productive discussion from happening, don't be surprised by this.

Let's all make sure that we're not above empathizing with the guy that invested $50000 in a rig that turned out not to be viable.

Your armchair psychology is really weird and inappropriate.

 

[markanini]

I've been following this thread quietly but have to speak out against remotely analyzing the psychology behind somebody's posts. Terms like 'Reactive Abuse' and attempts to read into past interpersonal relationships as impacting somebody's discourse on an audio forum is not appropriate.

I've found everyone's contributions interesting and some have taken considerable effort providing evidence and explanations. Disagreement exists but there's absolutely no need to read further into it than that.

The response evoked from you shows that you are driven by details in larger message. And less by the bigger message. The latter is not what you took the liberty to read into from my post evidently.

Which ties in with my main point, empathy doesn't come naturally to everyone, so it's good to let everyone know where your bias lies.

 

[Robbo99999]

Sorry, to be clear, I meant Andrew's thread on headphones.com - as you can see by my posting, I've been fairly active here.

From my POV, the thread on the forum serves as a little data point about preference, which points to the merits of at least testing @Resolve's pitch to let users adjust ear gain separately from the bass and treble. I'm not sure this testing will produce anything interesting - but I would have said much the same about Miller & Downey 2022's inquiry into a high-frequency peak filter, and so I'd have been wrong there!

(attached is the publicly available version Knowles put out)

Yep, Andrew's thread.

(It's not directly pertinent to this thread here, but I had a look at that publicly available Knowles pdf, on a personal level I wouldn't want the treble above 10kHz boosted as much as his target curve up there, but obviously I'm just one data point.)