He posted THD and IMD measurements (scroll back a couple of pages) so I'd assume yes
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Headphone Measurements Using Brüel & Kjær 5128 HATS
- Thread starter amirm
- Start date
You (@crinacle ) should post here more... good to see you around here
Discussions on ASR usually revolve around DACs, amps and sometimes speakers, which aren't really in my wheelhouse so I've never really had the opportunity to jump in.
Until recently
Andreas007
Active Member
I am going to sleep.
You should! I've noticed an increased amount of spelling errors.
IMO 41k is too much.
Especially, if we consider this fact:
Maybe a more affordable system can also deliver the most important values for a review site?
In the end, you're not going to develop ASR headphones, are you?
Last edited:
bidn
Active Member
Amir,
What you are doing is fantastic, as usual!
I enjoy so much your sharing all this trial and error on measuring headphone SQ!
I love so much this graph I quoted (first time I see this for headphones) of IMD vs dB, this is absolutely precious for me who listen at very low levels. I notice huge differences here, e.g. Focal headphones ( Elear, Clear, Elegia, Stellia, Utopia) or the Hifiman Arya are fine at very low levels while on the opposite the resolution of other headphones degrades too much (e.g. HE6se, Audeze 's LCDs) at these levels and require me to turn the volume up.
Just don't name it else you might struggle to give it back .
Oh and no sneaking off with it to cuddle in the camper van ! They definitely won't want it back after that.
Oh and no sneaking off with it to cuddle in the camper van ! They definitely won't want it back after that.
- Joined
- Nov 29, 2017
- Messages
- 584
- Likes
- 1,713
You should! I've noticed an increased amount of spelling errors.
IMO 41k is too much.
Especially, if we consider this fact:
Maybe a more affordable system can also deliver the most important values for a review site?
In the end, you're not going to develop ASR headphones, are you?
I'm not in opposition to the idea that $41k is too much - a GRAS fixture would be cheaper, for what it's worth, and might be worth trying out as well as they also do demos - but I'll note that the bolded section which I somehow missed earlier today is not correct. Nearest I can tell from a quick look at the spectrum of the output
the Audiocheck file is basically a log sine sweep that have been EQ'd to something like the Fletcher-Munson/equal loudness curves. The equal loudness contours are in turn derived from subjective level matching in anechoic and IIRC reverberant conditions - even setting aside the issues of subjective level matching with non-same perceived acoustic sources (I will not stop complaining about this, no), neither frontal free field or DF-HRTF are the Harman target. Even if you really buy into my historical argument that DF-HRTF is "close enough" to the Harman in-room baseline (which, for EQ purposes, it probably is to be fair), you still have the preferentially set shelving filters to contend with...and the SLD effect.
Be careful with jokes like that. I made one crack about that in 2017, and to this day people accuse me of romancing my 4128
restorer-john
Grand Contributor
restorer-john
Grand Contributor
eewww.
solderdude
Grand Contributor
Yes, Amir. This is something that is not seen elsewhere and can make a difference and is fun to explore. Perhaps a cheaper HATS can provide similar plots as well.
Bart
Member
Not incomparable to, perhaps, a chipped or broken tooth. The tongue continually scans the interior of the mouth creating a 3D 'picture' of it in the brain. If left untreated, after a while the chip is no longer noticed as an anomaly, but as an integral part of the 'picture'.
So, while the shape of our outer and inner ears may differ from person to person, do our brains callibrate in such a way as to allow us hear more or less the same things?
magicscreen
Senior Member
- Joined
- May 21, 2019
- Messages
- 300
- Likes
- 177
I do not understand why we need to compensate to the harman target curve?
- Joined
- Feb 23, 2016
- Messages
- 21,034
- Likes
- 38,292
Yes, excellent post here. Seems to me it is an unsolvable problem as it is being approached. All this kind of effort reminds me of epicycles in the Ptolemaic system. Though a flawed idea, various little refinements actually made it predict the planetary movements fairly well eventually. So much so that when the more correct Copernican explanation came along its biggest problem was that it lacked the precision of the old patched together Ptolemaic explanation.
This headphone problem is going to need something new and overlooked as an approach for it to actually work. Like Kepler's use of the ellipse in the Copernican system finally giving high accuracy in the correct explanation.
Wished I were the headphone Kepler, but I'm not. It is apparent the current approach isn't getting things too much further along according to my opinion.
I think the idea our brain continually recalibrates itself is supported by quite a bit of known information about how the brain works. One I've mentioned is within 2-3 weeks artificial pinna implants that totally ruin directional hearing are fully recalibrated to original accuracy. And when those are removed directional accuracy returns to a prior calibration even more quickly.
So what kind of testing can recalibrate on the fly in a way that it predicts headphones with the clearest most accurate response or do any of them? IEMs would seem to simplify matters because they bypass the pinna altogether. Still not an easy to solve problem even from there.
Awesome to see the testing beginning! I was initially confused about your "compensated graphs", but then I realised those were just showing the deviation from the Harman Target and therefore they should look nothing like the shape of the Harman Target Frequency Response Curve. I think I've seen a few members like @Fluffy say that those compensated curves don't look right and I believe people are not understanding that the compensated graphs are just showing the deviation from the Harman Target rather than the actual Frequency Response of the EQ'd headphone.....I also initially fell into this trap. EDIT: Although me personally I don't find the compensated graphs particularly useful or intuitive, to me just seeing the Raw Measured Frequency Response overlaid on the Target Frequency Response Curve is the most intuitive & useful presentation....but indeed that graph has been included too in the measurements, but I do think the compensated graphs are confusing people.As promised, here is the HD-650 results:
View attachment 78327
The two channels as measured are in red and green. They are pretty close to each other which is good.
We have to targets to compare it to. First one is in dashed blue and that is what @Mad_Economist has computed per intro. The other is BK supplied diffused field (dashed purple).
Re-measuring the same but having the Audio Precision software equalize the measurements using inverted Harman computed one we get:
View attachment 78328
This indicates too little bass below 100 Hz or so. There is also lack of energy between 2 and 6 kHz as noted.
Inversely, there is too much around 7 to 10 kHz. Using a much earlier measurement that varied somewhat from this, I made this EQ:
View attachment 78329
I was impressed by how much improvement it made. There was bass that I had never heard from HD-650 due to those two shelving filters. At normal listening levels, and across a couple of clips, I did not detect distortion either. Boosting though may cause digital clipping and there may be problems with other content.
The other two filters made impressive difference in bringing out detail and getting rid of bit of high frequency harshness (at least this is what I remember from a couple of days ago).
This is my confirmation method by the way on whether 5128 measurements are correct. If we act on them with equalization and the sound improves, then they must represent the truth.
Lots more experimentation is needed to determine the above but thought I share what I have for discussion. No, I have not tried Oratory EQ. Just no time. In general i am not a fan of many filters for equalization which he seems to be using. Real filter implementations have ringing and other artifacts that keep them from working as good as it seems on paper.
Anyway, the mission begins!
It's been interesting to see the distortion measurements of the different headphones so far and seeing the well renowned HD650 almost leading the field in this respect....another reason for it's good sound beyond it's frequency response.
I think @Mad_Economist 's work with providing the different computed Harman Target Curves using Paul Struck Method is fantastic, especially in light of the fact that we don't have an official Harman Target Curve for the B&K 5128....so I commend him on his creativity, knowledge, time & commitment to this project so far. In fact, I would be super stoked to actually listen to headphones EQ'd to his various Frequency Response Target Curves vs Oratory1990 Harman Curves done on GRAS measurement device......so folks here that are lucky enough to have headphones like the HD650 that have been measured in this thread by Amir can freely do some listening comparisons if they know how to EQ to different Target Curves.....me personally I'd use REW to create parametric filters to fit the different Target Curves, I have the ability & experience to do that as I've done similar thing with my AKG K702 months ago when I was messing around bodging together different hypothetical "measured" frequency response curves and EQ'ing them to the Harman Target Curve using REW. So far Amir has not measured any of my headphones so alas I can not audition these for myself. (Just gotta get NAD HP50 or Sennheiser HD600 measured here, if I'm lucky!
)
Last edited:
I don't exactly get what you are asking.
If you meant why do we need to compensate at all. Then yes we can live without compensation just to read the raw data. But you have to have a target in your mind at each frequency point. The reason why we need to have a target is because the difference between flat speakers and flat headphones are the HRTF. Basically what you actually hear at your eardrum is not what you measure at the 1m from speaker. To completely bypass the guess work you measure at eardrum reference point. The HATS is built for that. It simulates the behavior and impedance of human ear canal, pinna, head and torso. This way we can get a target that's relatively coherent with our speakers.
If you meant why Harman target instead of others. We don't. That's the point. Harman target was supposed to be the preferred response of headphones. But it's not necessarily what we want to see here.
Ha - thanks. You just mostly answered the question I was going to ask - why Harman? Their products don't suggest they know something that, for example, B&K don't?
Also - you refer to "flat" speakers and headphone - but none of them are very flat. Is that error more significant than HRTF, perhaps, as the human perceptual system already learns to compensate for HRTF?
Also - you refer to "flat" speakers and headphone - but none of them are very flat. Is that error more significant than HRTF, perhaps, as the human perceptual system already learns to compensate for HRTF?
uwotm8
Senior Member
- Joined
- Jul 14, 2020
- Messages
- 416
- Likes
- 486
Well I was using a "harmanizing" parametric EQ with some head- and earphones. In a month or so ended setting it off bc I tired of overcomfortable, overcompensated and a somewhat V-like dull lifeless sound. The whole idea of Harman curve looks/sounds more like the "best selling" sound, not the "fair' one. The good: you can listen to almost any music and it never gets harsh. The bad: it all sounds the same - same boring.
Sorry Harman, not buying this
Sorry Harman, not buying this
I do take your word for it, but I have to say that I am surprised that shelving up the bass would have made that great of a difference, without adding distortion. Good to know. That would suggest that the Roll off is caused by the electrical characteristic of the driver ratter than a physical limitation? Pardon my ignorance here, I'll admit my knowledge on this is limited and maybe I don't make sense, Is it common across the board to be able to fix the commonly seen bass roll of on open/dynamic cans by simply EQing those subs/lower bass back in? That would be great if you do dive in headphone measurments, to show post EQ measurments, Presently in my current rig I don't use digital attenuation (not because of a particular technical beliefs/principles) So yes in my case Digital clipping could be a concern, and my headphones have deeper extension than 650s, so never really went that route, but I have the curiosity.As promised, here is the HD-650 results:
View attachment 78327
The two channels as measured are in red and green. They are pretty close to each other which is good.
We have to targets to compare it to. First one is in dashed blue and that is what @Mad_Economist has computed per intro. The other is BK supplied diffused field (dashed purple).
Re-measuring the same but having the Audio Precision software equalize the measurements using inverted Harman computed one we get:
View attachment 78328
This indicates too little bass below 100 Hz or so. There is also lack of energy between 2 and 6 kHz as noted.
Inversely, there is too much around 7 to 10 kHz. Using a much earlier measurement that varied somewhat from this, I made this EQ:
View attachment 78329
I was impressed by how much improvement it made. There was bass that I had never heard from HD-650 due to those two shelving filters. At normal listening levels, and across a couple of clips, I did not detect distortion either. Boosting though may cause digital clipping and there may be problems with other content.
The other two filters made impressive difference in bringing out detail and getting rid of bit of high frequency harshness (at least this is what I remember from a couple of days ago).
This is my confirmation method by the way on whether 5128 measurements are correct. If we act on them with equalization and the sound improves, then they must represent the truth.
Lots more experimentation is needed to determine the above but thought I share what I have for discussion. No, I have not tried Oratory EQ. Just no time. In general i am not a fan of many filters for equalization which he seems to be using. Real filter implementations have ringing and other artifacts that keep them from working as good as it seems on paper.
Anyway, the mission begins!
Last edited:
OMG amir! This is awsome work! I'm so impressed!
Regarding the headphone positioning, is there a set of lines around the ears that you can use to line up the headphones so they are centered on the HATS?
Can I also suggest a heating element that is set to human body temp on the head, and that the phones are placed on there for a few mins before measurement?
Just my 2p
Otherwise, again.....
Regarding the headphone positioning, is there a set of lines around the ears that you can use to line up the headphones so they are centered on the HATS?
Can I also suggest a heating element that is set to human body temp on the head, and that the phones are placed on there for a few mins before measurement?
Just my 2p
Otherwise, again.....
Similar threads
