HBK Headphone Measurement Talks from Head-Fi and Sean Olive

[solderdude]

The big question is ... is only 1 rig correct in an absolute sense or are neither but one is closer to 'reality' and is this true for all headphones ?

 

[Deleted member 16543]

Hello,

Why the long winded response?

Because you keep missing the point.

 

[thewas]

The big question is ... is only 1 rig correct in an absolute sense or are neither but one is closer to 'reality' and is this true for all headphones ?

None will be ever absolutely "correct" for everyone, but from the design of the new B&K shouldn't it be closer to an average human ear?

And in that case someone should measure several of the best sounding headphones with it which should give a clue about its target curve.

 

[Deleted member 16543]

That is not supported by everyone, many people agree that above transition frequency you shouldn't correct direct sound based on listening position measurements if the loudspeaker has a flat direct sound and smooth directivity.

Some comments from Floyd Toole about room curve targets, room EQ and more

Room curve targets Every so often it is good to review what we know about room curves, targets, etc. Almost 50 years of double-blind listening tests have shown persuasively that listeners like loudspeakers with flat, smooth, anechoic on-axis and listening-window frequency responses. Those with...

www.audiosciencereview.com

My point is that flat DR and great directivity are not the ONLY way to get a balanced sound at the listening position.
And even if you had such an ideal speaker, given the acoustical peculiarities of each room, you can always go from pretty good to just right sound balance by correcting the whole audio range.

Room is to the speaker sound what outer ear (and leakage) is to headphones sound.

 

[Deleted member 16543]

The big question is ... is only 1 rig correct in an absolute sense or are neither but one is closer to 'reality' and is this true for all headphones ?

There's no such thing as a correct rig for everyone.
All there can be , and I think with the B&K 5128 there finally is, is an accurate average reproduction of the head and outer ear effects, which can help equalize headphones so that they will sound pleasurable to listen to the vast majority of the population. Even more people than Harman did manage to do.
The outliers will always be there, though.
It's just that with less anatomically accurate rigs there are more of them.

 

[thewas]

My point is that flat DR and great directivity are not the ONLY way to get a balanced sound at the listening position.
And even if you had such an ideal speaker, given the acoustical peculiarities of each room, you can always go from pretty good to just right sound balance by correcting the whole audio range.

It is the objective way, the rest is taste and audios circle of confusion.

Room is to the speaker sound what outer ear (and leakage) is to headphones sound.

They are not exactly comparable, listening to headphones adds an additional variation factor, which is the (partial circumvention of) individual HRTFs.

 

[solderdude]

There's no such thing as a correct rig for everyone.
All there can be , and I think with the B&K 5128 there finally is, is an accurate average reproduction of the head and outer ear effects, which can help equalize headphones so that they will sound pleasurable to listen to the vast majority of the population. Even more people than Harman did manage to do.
The outliers will always be there, though.
It's just that with less anatomically accurate rigs there are more of them.

Its unfortunate the EQ based on 5128 differs from that of the current usual suspect standards.
I wonder what this means in terms of reliability of the EQ based on these fixtures when the same tonal balance (room-ish target) is the ultimate goal.
In most cases though EQ loosely based on both methods common 'errors' may well be an idea but then perhaps leaning towards 5128 ?

 

[Deleted member 16543]

It is the objective way, the rest is taste and audios circle of confusion.

They are not exactly comparable, listening to headphones adds an additional variation factor, which is the (partial circumvention of) individual HRTFs.

Taste, by definition, cannot be accounted for. So all we can do is look at the objective side of the equation.

And yes, you're right about headphones. I thought I made my position aboundantly clear before, that they are even less predictable than speakers as far as final balance perceived by the listener goes.

 

[Deleted member 16543]

Its unfortunate the EQ based on 5128 differs from that of the current usual suspect standards.
I wonder what this means in terms of reliability of the EQ based on these fixtures when the same tonal balance (room-ish target) is the ultimate goal.
In most cases though EQ loosely based on both methods common 'errors' may well be an idea but then perhaps leaning towards 5128 ?

I don't think it's unfortunate. It's to be expected.
The only unfortunate thing is that B&K hasn't gotten off their couch and produced an official target response (room-ish based, as you call it) yet.

 

[solderdude]

Expected yes... but unfortunate that there has been so much EQ generated on so many different rigs that already differs and now there might be again another EQ which, by extension, should be more accurate.

I think for B&K the most important things are getting DF and FF corrections up to a standard which is relatively easy to do.
Put the 5128 in a calibrated room (for FF and DF) and play a sweep and there you have the curves for those standards.

If a 'standard' headphone would exist that always measures as it sounds to 'average' people then creating a curve (which will probably not be Harman) would be just as easy to do.
Unfortunately there is a lot of interaction and variables with headphones (on-ear, over-ear, in-ear, earbud, drivers dangling in free air in various angles before ears, the huge amount of available headphones etc that I can imagine this isn't an easy task.

It could be shortened by measuring only the best sounding ones... but best sounding to who ?

 

[Deleted member 16543]

It could be shortened by measuring only the best sounding ones... but best sounding to who ?

The person who owns the rig. There's no other way around it.
Enough people start doing that, and soon there should be enough data to point to a likely target, albeit unofficial, in lack of one provided by B&K.

The real problem is the high cost of the rig (ridiculously so, to be honest).
So it would have to come from the DIY community.
But how many DIY guys are there with anatomically accurate rigs?
I can tell you, from the units sold on my Shapeways shop, not that many.

 

[GaryH]

Hello All,

If you are most people your very own pinna and ear canal attached to your head are different than mine and different than the GRAS 45 CA-10 or B&K HATS 5128 test fixtures. Why would you want to introduce the bias and errors of an artificial pinna and coupler if a flat plane test fixture could perform better?

Chances are, even if you purchased a perfect Harman Target matching set of headphones, that you would if you could adjust the headphone equalization to match your own psychological preference and your very own pinna and ear canal acoustic peaks notches and dips.

Read Harman’s own, U. Horbach AES Convention Paper 9274.



U. Horbach, "Characterizing the Frequency Response of Headphones—A New Paradigm," Paper 9274, (2015 May.). doi:

Abstract: Traditional headphone measurements suffer from large variations if carried out on human subjects with probe microphones, and standardized couplers introduce additional biases, as concluded in a recent paper. Beyond that, there is no clear indication in literature about what the actual perceived frequency response of a headphone might be. This paper explores new measurement methods that avoid the human body as much as possible by measuring the headphone directly, in an attempt to overcome these restrictions and gain more accuracy. Design principles are described in the second part. A novel, DSP controlled, high-quality headphone is introduced that offers the ability to auto-calibrate its frequency response to the individual who is wearing it.



Thanks DT

My thoughts on the Harman Target curve:

The target Curve is a good middle of the road place to start (Least Sum of Squares), it is not too far away from the preference of most headphone users. After a couple of days of break in your new Harman Curve compliant headphones will be your new normal

The author of that paper, Dr Ulrich Horbach, was actually one of the chief engineers on the AKG N90Q, which is the headphone mentioned at the end of the abstract. I talked about this headphone and its auto-calibration ability via playing and recording an acoustic signal and its reflections by the ear in this post earlier in this thread. I've actually seen there are measurements of its response after calibration both on a human head and HATS, by Jude of Head-Fi:

As Jude says:

What we did for these measurements is put the AKG N90Q on someone's head and then used TruNote calibration to tune/correct the headphone's frequency response for that listener (a human). With that calibration in place, we put the AKG N90Q on the G.R.A.S. 45BB-12 KEMAR (a manikin with anthropometric pinnae) and measured it (dotted lines). Taking care not to move the headphones while still on the G.R.A.S. 45BB-12, I gently pressed the TruNote calibration button to then tune/correct the AKG N90Q for the 45BB-12's ears/head and then re-measured it (solid lines).

Now presumably this TruNote calibration must aim to tune to the same heard target response for every wearer (including the response 'heard' by a HATS). That means if you subtract the response calibrated on the KEMAR HATS above (solid line) from the response calibrated on-head (dotted line), you should get the 'PRTF correction curve' for that listener in relation to the KEMAR i.e. the difference in PRTF between the listener and HATS. What I was wondering, is maybe you could use this as a correction curve for that listener if convolved with the response of other headphones as measured on the same measuring rig. How could this theory be tested? EQ another headphone to the Harman target based on its standard measured response, then convolve this EQ with the listener's personal correction curve from the N90Q measurements as above to create a corrected personalised EQ. Then blind A/B test these two EQs to find out if the listener does indeed prefer the latter. All this could be done by anyone with an industry standard measuring rig (with the same acoustic impedance as the human ear) and an N90Q. Owners of N90Qs could even send in their pair pre-calibrated using TruNote to be measured in order to get their personalised PRTF correction curve (assuming the headphone saves the calibration while it's off during transit), which they could then apply to EQs for other headphones.

 

[Deleted member 16543]

The author of that paper, Dr Ulrich Horbach, was actually one of the chief engineers on the AKG N90Q, which is the headphone mentioned at the end of the abstract. I talked about this headphone and its auto-calibration ability via playing and recording an acoustic signal and its reflections by the ear in this post earlier in this thread. I've actually seen there are measurements of its response after calibration both on a human head and HATS, by Jude of Head-Fi:

As Jude says:


Now presumably this TruNote calibration must aim to tune to the same heard target response for every wearer (including the response 'heard' by a HATS). That means if you subtract the response calibrated on the KEMAR HATS above (solid line) from the response calibrated on-head (dotted line), you should get the 'PRTF correction curve' for that listener in relation to the KEMAR i.e. the difference in PRTF between the listener and HATS. What I was wondering, is maybe you could use this as a correction curve for that listener if convolved with the response of other headphones as measured on the same measuring rig. How could this theory be tested? EQ another headphone to the Harman target based on its standard measured response, then convolve this EQ with the listener's personal correction curve from the N90Q measurements as above to create a corrected personalised EQ. Then blind A/B test these two EQs to find out if the listener does indeed prefer the latter. All this could be done by anyone with an industry standard measuring rig (with the same acoustic impedance as the human ear) and an N90Q. Owners of N90Qs could even send in their pair pre-calibrated using TruNote to be measured in order to get their personalised PRTF correction curve (assuming the headphone saves the calibration while it's off during transit), which they could then apply to EQs for other headphones.

This would only work for N90Q headphones.
Different models will have different PRTFs.
How much those PRTFs differ between each other could indeed be interesting to know, though.

 

[MayaTlab]

That means if you subtract the response calibrated on the KEMAR HATS above (solid line) from the response calibrated on-head (dotted line), you should get the 'PRTF correction curve' for that listener in relation to the KEMAR i.e. the difference in PRTF between the listener and HATS. What I was wondering, is maybe you could use this as a correction curve for that listener if convolved with the response of other headphones as measured on the same measuring rig. How could this theory be tested? EQ another headphone to the Harman target based on its standard measured response, then convolve this EQ with the listener's personal correction curve from the N90Q measurements as above to create a corrected personalised EQ. Then blind A/B test these two EQs to find out if the listener does indeed prefer the latter. All this could be done by anyone with an industry standard measuring rig and an N90Q. Owners of N90Qs could even send in their pair pre-calibrated using TruNote to be measured in order to get their personalised PRTF correction curve (assuming the headphone saves the calibration while it's off during transit), which they could then apply to EQs for other headphones.

Unless I'm mistaken about the logic you're following, I think it wouldn't work for a simple reason : there isn't a constant transfer function between listeners or between listeners and HATS for all headphones. The HPTF is both listener / HATS and headphones dependent.
The difference you'd obtain would be specific to the N90Q, and dare I say maybe even to a specific sample of the N90Q if it's very sensitive to the state of its pads.

BTW, some of us are interested in knowing where exactly did you source these graphs "from Oratory" ?

https://imgur.com/Dq1vxPr

https://imgur.com/hHko7rE

 

[DualTriode]

Hello All,

In terms of a headphone target curve we hear a lot about accuracy and what target curve is potentially more accurate.

My though about target curve accuracy is that if your concern is a perfectly accurate curve your point of aim is off in the dirt somewhere.

My thoughts on the Harman Target curve:

The target Curve is a good middle of the road place to start (Least Sum of Squares), it is not too far away from the preference of most headphone users. After a couple of days of break in your new Harman Curve compliant headphones will be your new normal.

You will always be applying your personal preference and personal equalization to every set of headphones that you put on your head.

Dr. Horbach tells us that we need not measure SPL at our ear-drum with a probe microphone. Dr. Horbach also tells us that we do not need measure at the Drum Reference Point of an artificial ear either.

Dr. Horbach tells us that a measurement reference point inside the headphone cup measured on a flat plate test fixture is just as good as or better than a probe microphone at a real ear-drum or a DRP in an artificial ear. For the AES Conference Paper this Flat Plate test fixture used an electret microphone array.

The next step in refinement Dr. Horbach added a microphone array inside the cup of an AKG set of headphones with automatic machine calibration. Remember that this self calibration is part of the state of the art BRS (Binaural Room Scanning) 3D measurement system at Harman.

If you look at solderdude’s flat plate headphone plots they have very much the same shape and contour of Dr. Horbach AKG headphones with an electret microphone array.

Solderdude you could calculate a target curve to use with your database of flat plate headphone measurements.

Thanks DT

 

[Deleted member 16543]

Hello All,

In terms of a headphone target curve we hear a lot about accuracy and what target curve is potentially more accurate.

My though about target curve accuracy is that if your concern is a perfectly accurate curve your point of aim is off in the dirt somewhere.

My thoughts on the Harman Target curve:

The target Curve is a good middle of the road place to start (Least Sum of Squares), it is not too far away from the preference of most headphone users. After a couple of days of break in your new Harman Curve compliant headphones will be your new normal.

You will always be applying your personal preference and personal equalization to every set of headphones that you put on your head.

Dr. Horbach tells us that we need not measure SPL at our ear-drum with a probe microphone. Dr. Horbach also tells us that we do not need measure at the Drum Reference Point of an artificial ear either.

Dr. Horbach tells us that a measurement reference point inside the headphone cup measured on a flat plate test fixture is just as good as or better than a probe microphone at a real ear-drum or a DRP in an artificial ear. For the AES Conference Paper this Flat Plate test fixture used an electret microphone array.

The next step in refinement Dr. Horbach added a microphone array inside the cup of an AKG set of headphones with automatic machine calibration. Remember that this self calibration is part of the state of the art BRS (Binaural Room Scanning) 3D measurement system at Harman.

If you look at solderdude’s flat plate headphone plots they have very much the same shape and contour of Dr. Horbach AKG headphones with an electret microphone array.

Solderdude you could calculate a target curve to use with your database of flat plate headphone measurements.

Thanks DT

Hello,

Nobody cares about what you or Dr. Horbach think, as it has no correlation to real world scenarios.

Thank you,

Sax512

 

[solderdude]

Solderdude you could calculate a target curve

My target curve is a straight line in my plots.
It means (basically) a small bass boost and very slight drop-off in the treble.
This makes a 'horizontal line' slightly bass boosted (a'la equal loudness contours so not a steep bass filter as was used in the Harman determination of preference) which makes my plots appear somewhat rolled off in the sub lows compared to uncompensated flat plates.
I also have a slight roll-off in the treble so on the plots the treble appears higher than in some other plots.
Basically my plots look similar to how a flat speaker would measure at 1m in anechoic conditions.
Because the test fixture lacks a pinna the 1 to 5kHz region is 1 or 2 dB lower than with a pinna.
With angled drivers this may be more like 3 or 4 dB.

 

[DualTriode]

My target curve is a straight line in my plots.
It means (basically) a small bass boost and very slight drop-off in the treble.
This makes a 'horizontal line' slightly bass boosted (a'la equal loudness contours so not a steep bass filter as was used in the Harman determination of preference) which makes my plots appear somewhat rolled off in the sub lows compared to uncompensated flat plates.
I also have a slight roll-off in the treble so on the plots the treble appears higher than in some other plots.
Basically my plots look similar to how a flat speaker would measure at 1m in anechoic conditions.
Because the test fixture lacks a pinna the 1 to 5kHz region is 1 or 2 dB lower than with a pinna.
With angled drivers this may be more like 3 or 4 dB.

Solderdude,

Yes your test fixture plots do not suffer from all the pinna induced peaks dips and notches. Your test plots do not suffer from all the acoustic impedance interaction and resonances of a ear canal either.

I do this stuff for fun.

Solderdude I will send you money to purchase a JBL Tune 710BT set of headphones to measure. We can compare measurements from my flat plate test fixture and my GRAS CA 45-9. The purpose of this is fun and learning.

Please send me an email address where I can send the PayPal funds for a pair of JBL Tune 710BT.
I purchased the blue ones.

Thanks DT

 

[solderdude]

The purpose of this is fun and learning.

Fun, learning and comparing I'm game as it is for sharing let's share the costs as well after all it is only € 80.- when I decide to keep it (unlikely).
Looks to be half on-ear half over ear (very little room for ears)
Just ordered it (seems to have been a blue one as well).
Should be here in a few days.
Maybe next Thursday I have time to measure and evaluate.
When I will keep it I will PM you my PP address.

 

[DualTriode]

Hello,

Nobody cares about what you or Dr. Horbach think, as it has no correlation to real world scenarios.

Thank you,

Sax512

Hello All,

So there you have it, a strong opinion producing a strong emotional response without evidence to back it up.

Thanks DT