ASR Headphone Testing and BK 5128 Hats Measurement System

[solderdude]

I'm glad that we all seem to be (more or less) on the same page re what could potentially constitute a more ideal model for this type of a neutral response though.

I think everyone agrees about this. Unfortunately the answer has nothing to do with DF speaker measurements. You would have to understand how the DF correction is obtained in a non-reverberant room and the substantial difference this is with headphones.

The proposition @sax512 made to redo Harman research, perhaps in the same lab, is the only way to achieve a reasonable but still flawed method of obtaining a correction curve for that specific HATS.

I would not introduce preference (which is more of a commercial reason) in the target but try to find what an how much and why a bass compensation is needed.
I figure I would end up close to but not exactly the same as Harman research most likely and that would only be valid for that type of HATS.
Would want/need to use at least 10 or more different headphone types to try and find what exactly would be needed (an 'exacter' compensation).

Interesting but unaffordable and too time consuming to do this for the greater good and to publish it publicly.
Researches need to eat and live too and very expensive tests and research (using people) all costs money.

 

[ADU]

Fwiw, I agree with some (but not all) of what you're sayin above solderdude. Before we move on though, I had just a few more general thoughts re the subject of Harman target translation that I thought I should probably also share. So I'll post a few those now...

Although I have not read all the rest of this topic yet, I'm assuming that some other similar approachs to translating or correcting the Harman target for the 5128 rig have probably already been tried before here, maybe using the response curves of a couple different headphones for compensation (rather than using a diffuse or free sound field). And that these approaches were probably not entirely successful. Perhaps for some reasons similar to the ones mentioned in my last post above.

So I'm not going to bother trying to go down that particular path again. And will instead focus most of my attention on some other methods of deriving a neutral response curve for the 5128 rig, which are based more on some general approximations of the Harman curve's DF response, and the sound power responses of loudspeakers. Because I think those might stand a better chance of yielding some better results than the other translation approachs discussed in my previous post above.

I expect that there will be some similar objections to these other methods as well though. And look forward to hearing (and maybe also trying to respond to) some of those objections as well. (Some of which solderdude has already mentioned above.)

I'd have to agree with Amir (and others) though that any approach to simply deriving a neutral target that's done along these sort of lines would probably be somewhat limited in its applications,... unless or until some actual measurements can be done with a 5128 rig in a room with some actual loudspeakers to better verify their accuracy and legitimacy. I'll do my best though to try to explain and justify why I think a few of these other approachs might at least make a somewhat useful starting point or stop-gap for a neutral 5128 response in the meantime though. At least until some better in-ear data made in a room comes along.

 

[Deleted member 16543]

Interesting but unaffordable and too time consuming to do this for the greater good and to publish it publicly.
Researches need to eat and live too and very expensive tests and research (using people) all costs money.

That's why when I've been saying that somebody will eventually do this work, in my mind I've always been thinking about some university.
I guess we'll see if I'm right or not..

 

[ADU]

I think everyone agrees about this.

It is significant to me that there seems to be some general agreement on this here. Because this not exactly what I've encountered in some other audio circles.

If, for example, you were to put several reviewers or audiophiles into the same room together, and then ask them the same type of question of what represents a neutral or accurate or uncolored response, I sort of doubt that they would all have the same amount of clarity in their answers. (Maybe I'm wrong about that though, and selling some of these folks a bit short.)

The fact that we can all at least agree somewhat on what this is or should be though definitely gives me a little ray of hope.

 

[ADU]

This is beginning to get a bit off topic for this thread, Robbo99999. Which I know is something you don't like. But if you believe that the current Harman curve is accurate, then you should be able to take Amir's, or Oratory's, or Jaakko's, or someone else's Harman-based EQ corrections (whichever datasets you think are the most reliable), and then simply apply their recommended EQ settings to the 5128 measurements of the same headphones. And then compute the average frequency response of those 5128 headphone measurements with the EQ corrections applied to derive a similar sounding target response curve for the 5128 rig.

I believe that the Harman curve is not completely correct though. And can use some improvement in some spots, especially in the higher frequencies (as I've said on several occasions here). And believe that an approach like the one above would carry over those potential errors or inaccuracies to the new 5128 data, possibly making them even worse. So I prefer to use a different approach which is based more on Harman's process of estimating the sound power responses of loudspeakers (known as "spinorama"). And using that as my target for the 5128 diffuse field compensated measurements instead.

It may not be 100% accurate either. But at least it will not carry over any potential errors or inaccuracies in the current Harman curve. And would probably be quite a bit more accurate than the approach above (using the Harman EQ curves). So there's really nothing to be lost by using this other approach... It can't work any worse than using the current Harman curve as a model. And has the potential to be quite a bit more accurate than the Harman curve, particularly in the higher frequencies. So I don't really see where there's any downside to it. (?)

If you believe the current Harman curve is correct though, then the first approach discussed above of simply applying the current Harman EQ curves to the 5128 measurements should work just fine for deriving your 5128 target response curve. In addition to Amir's 5128 measurements, there are also some other measurements made by Jude for the 5128 rig in some of his recent headphone reviews, and also a couple more in his Head-Fi Audio Measurements Lab that you could use to begin this process.

Using sound power is a more accurate way to go though, imho. With fewer potential downsides, and alot more potential upside!

Hadn't noticed this before, but the SoundGuys now also have some headphone measurements made on their 5128 rig. These only date back to about May of this year. And you can tell which are the new plots made on the 5128 by the color of the background, which is a darker gray. (Rather than the white background they were using before.) Fwiw, they have recent 5128 measurements for the following fairly well-known over-ear HPs, and probably also a few other consumer brands (though I didn't bother looking at those).

AKG K371
Audeze Mobius
AudioTechnica M40X
Beyerdynamic DT-770 (80-ohm)
Beyerdynamic DT-880 (250-ohm)
Bose NC 700
Bose QC 35-II
Sennheiser HD 280 Pro
Sennheiser HD 6XX

They're using a mish-mash of different target response curves for these HPs (based, it appears, largely on some guesswork on their part), even though they are all over-ear HPs. So I wouldn't pay a whole lot of attention to those.

Since I don't have any more updates yet based on sound power, here are couple of my rough attempts at defining a 5128 target from a month or so ago. Which are based on diffuse field slopes that roughly approximate a well-extended sound power curve and the Harman curve...

These might be just a bit too elevated in the 2 kHz range. And the 2nd graph might also be a bit too elevated at the very ends of the curve in the sub-bass and very high treble. But they're probably in the general ballpark, if you need some kind of a rough reference for where a neutral response might be on the above graphs.

I am not really familiar with the SoundGuys site, so I don't know how reliable it, or their measurements are. I believe the diffuse field graph posted on their page explaining their 5128 target curves is actually for the GRAS KEMAR though. So maybe they were using a GRAS rig before they got the 5128, and just didn't realize there'd be a difference in how the two would measure in a diffuse sound field (which is a very common mistake).

 

[ADU]

A more recent version of my 5128 approximation curve, made from some slightly better modeling of loudspeaker sound power response data than the previous ones above.

This is still very much a work in progress which hasn't been tested with any headphone gear yet. And like the Harman target, it is also only useful for some very general approximations, esp. in the treble.

The peaks in a headphone's response at around 3, 8 and 16 kHz should probably follow the general shape of the above curve. But there should probably also be some noticeable dips or valleys that drop somewhat below this curve in the treble in between those peaks, for a more neutral response. When I get a chance, I'll try to show a few better examples of how I think the responses of some actual headphones should look compared to the above curve, to hopefully illustrate this a bit better. (I think most of you probably already get how this works though.)

The closest match that I've found to the above response curve so far is an early version of the SoundGuys 5128 target that was based mostly on averaging the responses of some neutral sounding headphone measurements together...

https://mk0soundguyshosprmrt.kinstacdn.com/wp-content/uploads/2021/07/sg-calc-jpg-1024x672.webp

The similarities in the overall shape of these two curves suggests to me though that the above approximation curve is probably beginning to get fairly close to the general ballpark.

This is all based just on using the 5128's diffuse field compensation curve, combined with the average sound power responses of a variety of neutral loudspeakers with a flat direct response btw. And my interpretations of that data. Rather than Harman's previous work with the GRAS system (which is not as relevant to the new 5128 system imho).

 

[ADU]

Some sound power data used for modeling the response curve above. These represent the average sound power responses for several different groups and quantities of loudspeakers with a flat direct response. But with the downward slope of the SP curves removed, so I could see how they tend to deviate from a constant slope a little better...

The frequency ranges shown on the above graphs are not consistent btw, so that's why I removed the dB levels from the graphs. Here are the descriptions and frequency ranges though for the 5 graphs...

My attempt at a crude approximation of some of these deviations from a constant slope. With the falloff in the bass extended down to 20 Hz...

This adjustment curve, combined with a -1.26 dB per octave slope, and the 5128 diffuse field compensation curve produced the new 5128 target approximation curve shown in my last post above.

 

[FireLion]

When will China create a full clone on Aliexpress?

Also I read somewhere that this $50k unit would monopolize measurements. Thoughts? Not sure just putting it out there.

Also how does something with less detail than a clothes shop window mannequin represent a real human?(I dumb).

If you are just measuring headphones and not speakers how does the $40k dummy tosro get utilized when it comes HRTF? (I da dummy, I should probably read the papers). IEMs to an even lesser extent.

When I logically think about it there is jut a small areas being covered even with the largest of headphones. A case might be made for open back, but again wouldn't the setup have a minimal effect on the result. Speakers, that's a different story.

Will the 711 clones still be the top device for enthusiasts when it comes to measuring?

I think this would make for a great video discussion or presentation, from those who are way more knowledgeable on the topic.

 

[617]

When will China create a full clone on Aliexpress?

Also I read somewhere that this $50k unit would monopolize measurements. Thoughts? Not sure just putting it out there.

Also how does something with less detail than a clothes shop window mannequin represent a real human?(I dumb).

If you are just measuring headphones and not speakers how does the $40k dummy tosro get utilized when it comes HRTF? (I da dummy, I should probably read the papers). IEMs to an even lesser extent.

When I logically think about it there is jut a small areas being covered even with the largest of headphones. A case might be made for open back, but again wouldn't the effect have a minimal effect on the result. Speakers, that's a different story.

Will the 711 clones still be the top device for enthusiasts when it comes to measuring?

I think this would make for a great video discussion or presentation with those who are way more knowledgeable on the topic.

With metrology you aren't paying for the equipment as much as the manufacturer's guarantee that the equipment is accurate, within spec, and traceable to some kind of standard.

 

[FireLion]

With metrology you aren't paying for the equipment as much as the manufacturer's guarantee that the equipment is accurate, within spec, and traceable to some kind of standard.

I should have looked it up on the B&K ear canal difference for HRTF.

Still a interesting topic and seems to be gaining a little momentum. Sean Olive has making a number of appearances on YouTube of late.