Headphone Measurements Using Brüel & Kjær 5128 HATS

[zepplock]

... They have better bass extension and bass quality, but much less quantity, definitely not in the same level with the mids. ...

Its a really simple drawing, but i would describre the perceived bass+mid like this:

can you explain extension, quality and quantity in measurable units?

 

[dasdoing]

Because if you would have a perfectly flat headphone and Measure it with a perfectly flat microphone at your eardrum you won't get a flat Measurement. The difference between what you would Measure and perfectly flat is the compensation. What is flat and stuff is another question of course but just to illustrate why a compensation is required.

sorry, I still don't get it.

let's simplify and say our ears just add a 10dB boost to 10k. a flat sounding headphone would therefore have a 10dB dip at 10k. meassuring that headphone with a perfect replica of our ears should then result in a flat curve, because it would also boost 10k with 10dB

 

[Mad_Economist]

sorry, I still don't get it.

let's simplify and say our ears just add a 10dB boost to 10k. a flat sounding headphone would therefore have a 10dB dip at 10k. meassuring that headphone with a perfect replica of our ears should then result in a flat curve, because it would also boost 10k with 10dB

The transfer function of the ear is not static, it depends on the sound field.

The delta between the response at a microphone without the pinna and canal sim and the response at the microphone with them is not what would need to be applied to the eardrum response to achieve perceived flat frequency response.

 

[dasdoing]

The transfer function of the ear is not static, it depends on the sound field.

The delta between the response at a microphone without the pinna and canal sim and the response at the microphone with them is not what would need to be applied to the eardrum response to achieve perceived flat frequency response.

So why pay $40.000 for a simulation that doesn't simulate what it tries to simulate?

 

[Mad_Economist]

So why pay $40.000 for a simulation that doesn't simulate what it tries to simulate?

The 5128 does simulate what it tries to simulate, and can reasonably claim to do so better than any other ear simulator.

 

[617]

Good Lord this is complicated

 

[Mad_Economist]

Good Lord this is complicated

From my standpoint, other audio measurements are ludicrously insufficiently complex

 

[617]

From my standpoint, other audio measurements are ludicrously insufficiently complex

Frankly I'd be satisfied if measurements showed narrowband breakup and other resonances. I really do believe a lot of these headphones show poor engineering. Just a hunch.

 

[Mad_Economist]

Frankly I'd be satisfied if measurements showed narrowband breakup and other resonances. I really do believe a lot of these headphones show poor engineering. Just a hunch.

I probably wouldn't put those that high on my list of headphone engineering sins, honestly, but I'd strongly affirm that there's a shockingly large number of headphones - including those released recently - that seem shockingly backwards or inexplicably designed. Indeed, if anything so much so that a really substantial, frequency-response disruption resonance in a driver doesn't necessarily knock a headphone into a very low performance tier...

 

[Blumlein 88]

Good Lord this is complicated

So much so that no one actually can make it work so far.

For instance the worlds greatest listener using a headphone that performs the best he has ever heard can suggest you try it on yourself. But with different HRTF it may literally perform like shite for you. Que' sera, sera.

 

[617]

So much so that no one actually can make it work so far.

For instance the worlds greatest listener using a headphone that performs the best he has ever heard can suggest you try it on yourself. But with different HRTF it may literally perform like shite for you. Que' sera, sera.

When you refer to HRTF are you referring to the inner ear resonances? I've only heard that term refer to the broader frequency response deviations caused by the head being between the ears.

Regarding the latter, is there a big audible difference between different people's heads? Or are the inner ear differences more prominent?

Im starting to see headphone engineering much the same way I see speaker engineering; the goal is not good sound in an individual ear but relative insensitivity to different ear shapes. I have no idea if anyone is succeeding in that goal.

 

[Mad_Economist]

So much so that no one actually can make it work so far.

For instance the worlds greatest listener using a headphone that performs the best he has ever heard can suggest you try it on yourself. But with different HRTF it may literally perform like shite for you. Que' sera, sera.

This is an overstatement of the variation, though. See this comment by @pozz and my comment adjacent, last paragraph.

 

[Mad_Economist]

Regarding the latter, is there a big audible difference between different people's heads? Or are the inner ear differences more prominent?

Depends - directional HRTFs at the canal entrance are probably a bit more variable than canal entrance > eardrum transfer functions between individuals, but the specifics of the canal resonances tend to vary more than the spreads of occluded canal DF-HRTFs I've seen.

Im starting to see headphone engineering much the same way I see speaker engineering; the goal is not good sound in an individual ear but relative insensitivity to different ear shapes. I have no idea if anyone is succeeding in that goal.

I wouldn't say that this is the goal, but I would definitely agree that it is a goal, and IMO likely to contribute meaningfully - if not the extent @Blumlein 88 is implying - to the variation in people's perceptions of headphones.

 

[Blumlein 88]

This is an overstatement of the variation, though. See this comment by @pozz and my comment adjacent, last paragraph.

I'd look if I was sure of which Olive source is used.

In chart from pozz's post, is that not one where they used a single headphone and altered the response to simulate different phones they had measured? Seems that would make for quite a different idea on how widely applicable that is. Or do I have the wrong paper by Olive?

 

[Mad_Economist]

I'd look if I was sure of which Olive source is used.

I typically do link these things, but it's A Statistical Model that Predicts Listeners' Preference Ratings of In-Ear Headphones, parts 1 and 2 that I'm referencing. Bear in mind, we're talking about a consistent clustering of preferences with response that, at most, can be individually modified by a small portion of the canal's length; essentially the worst-case scenario, from a standpoint that HRTF differences are the major timbral issue with headphones.

In chart from pozz's post, is that not one where they used a single headphone and altered the response to simulate different phones they had measured? Seems that would make for quite a different idea on how widely applicable that is. Or do I have the wrong paper by Olive?

That was the methodology used in the paper @pozz is referencing - am I misunderstanding your contention, here? I was parsing your complaint to be that without knowing the correlation of a headphone's in situ response and the wearer's HRTF, we can't predict subjective response.

 

[richard12511]

The 5128 does simulate what it tries to simulate, and can reasonably claim to do so better than any other ear simulator.

Hmmm, if it really does simulate what it claims to simulate, then I would expect some of these compensated curves to be at least close to a flat line. Is that thinking wrong? Either that or all of these headphones measured so far are beyond terrible(comparable or worse than the worst speaker measured). How do we know which is right? I'm kinda in the middle right now. Not sure what I believe, yet.

Honestly, if it can do what it says, then I don't think $40,000 is too steep of an asking price. Companies could make that back easily from extra sales generated by using it to design new headphones.

 

[pozz]

I typically do link these things, but it's A Statistical Model that Predicts Listeners' Preference Ratings of In-Ear Headphones, parts 1 and 2 that I'm referencing. Bear in mind, we're talking about a consistent clustering of preferences with response that, at most, can be individually modified by a small portion of the canal's length; essentially the worst-case scenario, from a standpoint that HRTF differences are the major timbral issue with headphones.


That was the methodology used in the paper @pozz is referencing - am I misunderstanding your contention, here? I was parsing your complaint to be that without knowing the correlation of a headphone's in situ response and the wearer's HRTF, we can't predict subjective response.

@Blumlein 88 The graph is from: https://secure.aes.org/forum/pubs/conventions/?elib=17500

 

[Francis Vaughan]

One thing about the $40,000 HATS is that it is a redundant setup. Paying for two ears when one will do. It isn't clear how much the full torso model is helping either, but that is likely of inconsequential cost.
A big part of what you pay for is repeatability. If you own one HATS, you expect the measurements to be identical on another. One could arge that that is less important here at ASR, but critical in an R&D or QA environment. A big part of what you get with the B&K logo is that repeatability.

 

[Mad_Economist]

Hmmm, if it really does simulate what it claims to simulate, then I would expect some of these compensated curves to be at least close to a flat line. Is that thinking wrong? Either that or all of these headphones measured so far are beyond terrible(comparable or worse than the worst speaker measured).

The expectation is the trouble, yes. Headphone frequency response is vastly more varied from any target than anechoic on-axis speaker response. Indeed, that's been kind of a meme for a while.

 

[Gomjab]

For peaks, 90 dB is not that loud particularly for highly dynamic material such as classical, jazz or other predominantly acoustic music. For example, using a pair of HD600, I can listen to the below without any discomfort with 99 dB peaks and an average level of 80 dB. Even for very compressed modern music (e.g. Muse), 90+ dB peaks don't bother me.

Wow that Afro Cuban All Stars link became the soundtrack for this evening’s web browsing! Very nice.