Harman preference curve for headphones - am I the only one that doesn't like this curve?

[Robbo99999]

Now we are talking about FR but somehow some headphones sound less bright despite having more treble or less boomy/present despite having more bass. I wonder if this is related to presentation and soundstage. On paper you see more treble but it's perceived as less treble with different location/distance by brain. That more treble gives different cues to the brain(to make sense of the sound) so you perceive it as less offensive.


View attachment 191455
An example in my case: Hifiman Sundara, pretty much harman tuned headphone with 90+ adherence score yet it sounds way brighter than DT 770 which is famously bright and peaky headphone. 770 sounds somewhat dull and diffused, distant, like sound comes from other room(kinda). Sundara on the other hand, sound raw and incisive, very forward and close to eardrum.
Example song:

Sundara sound disturbingly bright here. Splashes tearing my eardrums. 770 don't.


Or another example:
View attachment 191454
Arya is famous with it's bass. HD800S famous with it's lack of bass. Treble too, HD800S regarded as bright Arya seem brighter in graph.

Fit, comfort and sound presentation is crucial at making sense of the sound via brain. I feel our interpretation of FR is limited or FR is just one part of the puzzle.

From the first frequency response graph I could probably understand why Sundara would sound brighter than DT770, because you'll see that the DT770 has significantly more bass which makes it sound warmer. The Sundara also has more energy in the 2.5 to 4.5kHz area which is a place of high hearing sensitivity so increases in dB here will be readily felt/heard - so again that is an aspect that increases the brightness of the Sundara. So to me the frequency response graphs would support your listening observations of Sundara being brighter than DT770.

The second frequency response graph, I'd agree that you'd expect Arya to sound brighter. I don't know how many units of headphone were measured by Crinacle to come up with those averages, but if it's only based off one headphone for example, then the unit to unit variation factor could creep in, thereby making it hard for us to place too much weight on this comparison, we may not know if this reflects the reality of most of these headphones out in the wild. Having said that, the HD800s apparently has super low unit to unit variation - I heard Oratory say as much, I think he's measured a number of these HD800s......however I don't know what the unit to unit variation is like for the Arya, could be that this headphone is not being totally accurately portrayed in this frequency response graph.

So in conclusion, I think it's a bit hard to say that there are other factors that describe tonality apart from the measured frequency response of a headphone - ie your "presentation & soundstage" influence theory. I mean it's possible that some people don't get a good seal when they wear a certain model of headphone which would result in less bass and brighter sound vs published/measured frequency responses, so that's certainly a factor in differences.....but I don't really see the "soundstage" aspect of headphones "upsetting"/invalidating the tonality we see in measured frequency responses. I do think soundstage exists as a quality in headphones but I don't think it would have the influence you're suggesting, that's my read on it.

 

[markanini]

Now we are talking about FR but somehow some headphones sound less bright despite having more treble or less boomy/present despite having more bass. I wonder if this is related to presentation and soundstage. On paper you see more treble but it's perceived as less treble with different location/distance by brain. That more treble gives different cues to the brain(to make sense of the sound) so you perceive it as less offensive.


View attachment 191455
An example in my case: Hifiman Sundara, pretty much harman tuned headphone with 90+ adherence score yet it sounds way brighter than DT 770 which is famously bright and peaky headphone. 770 sounds somewhat dull and diffused, distant, like sound comes from other room(kinda). Sundara on the other hand, sound raw and incisive, very forward and close to eardrum.
Example song:

Sundara sound disturbingly bright here. Splashes tearing my eardrums. 770 don't.


Or another example:
View attachment 191454
Arya is famous with it's bass. HD800S famous with it's lack of bass. Treble too, HD800S regarded as bright Arya seem brighter in graph.

Fit, comfort and sound presentation is crucial at making sense of the sound via brain. I feel our interpretation of FR is limited or FR is just one part of the puzzle.

I don't see any contradiction in the first comparison. The DT770 has a dip at 4kHz and elevation below 150Hz that's gonna mask part of the treble. Such effects are well studied and form the base for mp3 compression: https://en.wikipedia.org/wiki/Auditory_masking

 

[Deleted member 23424]

Wow, this is a lot like Paul Barton’s methodology behind his RoomFeel tech. I heard the two targets were very similar but didn’t realise they used such similar methods.

On the “ASR should aim for accuracy” point, wasn’t it discovered that accuracy didn’t necessarily result in the most preferred sound for the majority of people, and the 2018 target reflects that? No use in recommending headphones that are “accurate” when “accurate, but tuned to be appealing to the greatest number of people” is available.

I personally agree with you and absolutely see the value in accuracy, but I don’t think everyone wants to adjust to accuracy. Most people probably want something that sounds good and don’t give a hoot whether it sounds like great speakers in a semi-reflective room.

Give me deviation/compliance to accuracy and potential EQ distortion data over how close a headphone sounds to somebody else's curve any day of the week.

 

[WickedInsignia]

However, I do think the 2018 Harman Curve should be used on this site as it is the response that Harman found most people liked, which should also equate on some level to a most accurate/authentic experience. That last point being that the extra bass is there to probably account for the fact you don't get tactile bass in headphones, and the added treble is probably there to balance out the added bass to overcome the masking effect of the increased bass.

I didn't understand the justification for 2018's deviations from the 2013 target well enough, thanks for the clarification there! This is a good case for ASR's target choice.

I really should go and read Harman's papers.
EDIT: I realize now that the papers are a serious time investment and somewhat inaccessible. The article at the link posted by NTK below looks like a great alternative!

 

[NTK]

After almost 4 years of no new post, Dr @Sean Olive just posted (March 4) a new one at his Audio Musings blog. He has an article published in the Spring 2022 edition of the Acoustics Today magazine -- a summary of the headphone research at Harman of the past decade. Link to the AT article is in the blog post.

Audio Musings by Sean Olive

A blog about the science of sound recording and reproduction

seanolive.blogspot.com

 

[ADU]

After almost 4 years of no new post, Dr @Sean Olive just posted (March 4) a new one at his Audio Musings blog. He has an article published in the Spring 2022 edition of the Acoustics Today magazine -- a summary of the headphone research at Harman of the past decade. Link to the AT article is in the blog post.

Audio Musings by Sean Olive

A blog about the science of sound recording and reproduction

seanolive.blogspot.com

Very nice. Thanks for posting this, NTK.

 

[ADU]

Even if you really, really like the so called Harman preference curve at let's say 75 dB when you crunch volume up 12 dB it will be very offensive (more so as most drivers will start to distort severe as you cross the +6 dB in bass region).
On the other hand if you do the same with relatively flat ones in bass region (crunch it up to 87 dB) you will get something similar to Harman curve but they won't distort that much.
Conclusion Harman or any other static curve is absolutely wrong way to go. Good old equal normalization curve (ISO 226 mesured) for given driver's at the desired listening level (dB) should be your starting point and work it out from there to what ever preference you wish.
And no funny business of emulating an emulation of emulation real given driver and their physical property's (mesured).

Not sure that I understand what you're suggesting here, Zolalll.

 

[ADU]

Just as a very general point - I have indeed noticed that the Revels don’t go all the way down to 20hz, but they still go very low.

Yes.

A simple ‘bass’ control is most certainly not, by any stretch of the imagination, an accurate substitute, and I don’t feel it’s an appropriate presumption to imply otherwise. Harman could have tweaked the EQ on the Revels, or added their own sub.

Probably true.

 

[ADU]

The 2013 Harman Curve is developed to track speakers that are extended all the way down to 20Hz.

Yes.

As I replied to you before it's exactly equal to the Harman Speaker Curve tilt applied to the in-ear dummy head measurement, that's a fact.....and the Harman Speaker Curve is extended all the way down to 20Hz (almost), in fact with a bass hump too, so there's plenty of bass there in terms of "frequency extension", it's more than linear extension.

So it's the tilt of the following dark black solid line...View attachment 191421, applied to the in-ear Harman measurement here View attachment 191426. At which point you then end up with the 2013 Harman Curve here which is the blue line View attachment 191427, note how that's extended down to 20Hz.

I haven't done the necessary math and plotting to confirm the above recently. But I'll accept your analysis for now that the 2013 Harman target is likely based on their measurement of the Revel F208 after it had been EQ'd to match as well as possible their PIRL curve at that time (shown in the graph below on the right), and their in-ear measurement of speakers EQ'd to a flat response at the listening position.

If you look at my previous comments above though, Robbo, you'll notice I said that both the current 2018 Harman target, and also Harman's preferred in-room loudspeaker curve (or PIRL) shown above appear to be more or less consistent with neutral speakers that are well-extended into the sub-bass, or accompanied with a sub-woofer, in terms of their bass response... And I still stand by that. And believe Dr. Olive also agrees with this, based on his previous comments. So I guess I don't understand the point that you're trying to make on this.

If your assumptions above about the 2013 target are correct though, then both curves (iow, not just the 2018 target) appear to be based to some extent on subjective preferences. Which seems to conflict with your and Yorkshire Mouth's characterizations of one being "accurate", and the other being "accurate+tweaked for preference". No?

 

[Robbo99999]

Yes.



I haven't done the necessary math and plotting to confirm the above recently. But I'll accept your analysis for now that the 2013 Harman target is likely based on their measurement of the Revel F208 after it had been EQ'd to match as well as possible their PIRL curve at that time (shown in the graph below on the right), and their in-ear measurement of speakers EQ'd to a flat response at the listening position.

If you look at my previous comments above though, Robbo, you'll notice I said that both the current 2018 Harman target, and also Harman's preferred in-room loudspeaker curve (or PIRL) shown above appear to be more or less consistent with neutral speakers that are well-extended into the sub-bass, or accompanied with a sub-woofer, in terms of their bass response... And I still stand by that. And believe Dr. Olive also agrees with this, based on his previous comments. So I guess I don't understand the point that you're trying to make on this.

If your assumptions above about the 2013 target are correct though, then both curves (iow, not just the 2018 target) appear to be based to some extent on subjective preferences. Which seems to conflict with your and Yorkshire Mouth's characterizations of one being "accurate", and the other being "accurate+tweaked for preference". No?

The 2013 Harman Curve is essentially that Preferred In-Room Target slope you mentioned applied to the in-ear measurement, however is has that dip at 200Hz which is the gap you see between the dotted red line (Preferred In-Room Target) and the solid black line (equalised Revel F208), so it's closer to the actual Equalised Revel F208 applied to the in-ear measurement.....however the 2013 Harman Curve does have the bass extention of PIR Target and the "straight line smoothing" above 1000Hz. So for simplification in my prior post I just said it was the F208 slope applied to the in-ear measurement, but actually it's a mixture of both, but we're talking small differences here. The conclusion to be made though is that 2013 Harman Curve is a direct translation of their Harman Speaker Curve applied to the in-ear measurement, so is the most accurate target if you think the Harman Speaker Curve is an accurate target for a loudspeaker in a room. For instance my anechoically EQ'd flat JBL 308p speakers follow the general trend of the Harman Curve even without any room EQ, so the Harman Speaker Curve is pretty much how anechoically flat speakers behave in a room.....and anechoically flat speakers being the agreed measure of perfection here on ASR.

To be clear though as I've said before, I think the 2018 Harman Headphone Curve that is used as the target here on ASR is the best curve to use for measurements on this site for the reasons I mentioned in my earlier post (https://www.audiosciencereview.com/...hat-doesnt-like-this-curve.19668/post-1113506 ).

 

[ZolaIII]

Not sure that I understand what you're suggesting here, Zolalll.

Take a look at ISO 226 graph for example for one of my favourite IEM's.

https://reference-audio-analyzer.pro/en/report/hp/creative-aurvana-trio.php#gsc.tab=0

Those aren't on Harman at all as mesured on 72 dB and main driver is paper cone. It's pretty much self explained. There is no unified stand what's a normal level on which you will listen nor on which it should be mesured nor it can be. Some people like quieter, other don't, some have to and other don't and even sometimes how we feel will determine how loud we will listen to (along with many other reasons). Driver response will change depending on that (loudness) and it will be different for different drivers along with distortion level's among other things.
They (Olive and others) used emulated "perfect" loudspeaker to calculate response on hedaphones which should be same and emulated all other hedaphones on single AKG model as I remember doing the research. This is wrong on so many fronts that I don't even want to begin explaining.
Simply use mesured - calculated equal loudness curves for given hedaphones and drivers at the loudness level you intend to listen on and EQ from there to what ever you like (as much as the driver allows you to).
As soon as you realise response is not fixed but depending on such factors regarding main one - loudness you realise how silly is all the fuss and discussion about supposed target curves, and more you think about it you realise they re invented hot water, wrapped it up nicely and sold it to you but as usual it doesn't work (as they disregarded individual driver characteristics and that loudness won't ever be on only one fixed level).

I hope that helps and that will end such discussions in the future.

 

[JustEnjoy]

Hi all

With some incredible help from fellow forum members here (@pkane , @Robbo99999 , @dasdoing ) I've been able EQ my headphones to the Harman preference curve.

I started with my own favourite Focal Elegia and just today I've EQ'd my friends Aeon 2 Closed and HD800-S , all using oratory1990 EQ parameters to achieve Harman curve.

With all these headphones I really hate the Harman curve (please don't delete my account @amirm )

It sounds nothing like my nearfield desktop Genelec monitors (1m from my head).

I thought maybe it was my Focal Elegia but I have the same tonality when I eq Aeon 2 closed and HD800-S

A flat headphones curve sounds closer to the nearfield speakers to me.

It's possible that as I continue to tweak target curves and listen, that I end up preferring something between these 2 extremes of flat vs Harman curve - the journey continues, so please don't shoot.

Am I alone here in really not liking Harman preference curve? Like really not liking it

Why does anyone have to tell us what we should like, can’t we decide for ourselves. When I listen to different headphones that are designed to the Harman curve I feel as if they are flat and dull. I have attended hundreds of live performances in my life and they never sound anything like a headphone tuned to the Harman curve, voices are lively, horns and percussion pop, the music is fun. But we each get to decide for ourselves what we like, I say stop arguing and just enjoy the music…

 

[RHO]

have attended hundreds of live performances in my life and they never sound anything like a headphone tuned to the Harman curve, voices are lively, horns and percussion pop, the music is fun.

I suspect that the live performance is a bit louder than when you listen to music using headphones.
Try playing you headphones as loud as the live performance. (If they can handle it)

 

[watchnerd]

I suspect that the live performance is a bit louder than when you listen to music using headphones.
Try playing you headphones as loud as the live performance. (If they can handle it)

Isn't one of Amir's headphone tests at 115 dB?

 

[majingotan]

Why does anyone have to tell us what we should like, can’t we decide for ourselves. When I listen to different headphones that are designed to the Harman curve I feel as if they are flat and dull. I have attended hundreds of live performances in my life and they never sound anything like a headphone tuned to the Harman curve, voices are lively, horns and percussion pop, the music is fun. But we each get to decide for ourselves what we like, I say stop arguing and just enjoy the music…

What you see on the FR curve doesn't exactly translate to what arrives to your eardrums and your ear anatomy is unique to you such that the resonances from the sound pressure (mids and treble are more prone to this) is slightly different from person to person, and don't forget your preferences in bass response as well. Just looking at the Dan Clark thread here showed polarizing impressions of a headphone targeting the Harman Curve carefully. For IEMs, my ear preference is inverse Harman where I hated excessive pinna gain and muted treble and thin upper bass of the Harman IE curve hence I chose an IEM that is inverse of that (that's the IEM in my profile pic). For headphones, Harman Curve OE is pretty much "tonally correct" to my subjective preferences.

 

[markanini]

So did anyone that felt dissatisfied get to the bottom of whether the Harman curve was to blame, in any of the following ways:

• Try a natively Harman tuned headphone to rule out bad EQ.
• Listen to your headphones without EQ, some are already close to Harman in a stock state, so further EQ might actually be reducing accuracy due to inter unit variation and/or Personal anatomical variation.
• Ease off on low end EQ, as it can drive headphones into distortion many times, and should be adjusted to taste anyway:

With a total lack of follow through if someone wanted to make the point that the critique exists purely for being contrarian it would be warranted.

 

[m8o]

So did anyone that felt dissatisfied get to the bottom of whether the Harman curve was to blame, in any of the following ways:

• Try a natively Harman tuned headphone to rule out bad EQ.
• Listen to your headphones without EQ, some are already close to Harman in a stock state, so further EQ might actually be reducing accuracy due to inter unit variation and/or Personal anatomical variation.
• Ease off on low end EQ, as it can drive headphones into distortion many times, and should be adjusted to taste anyway:

With a total lack of follow through if someone wanted to make the point that the critique exists purely for being contrarian it would be warranted.

I have not bought lesser headphones that track close to the target. But, I believe the DCA Stealth is considered to have excellent tracking to the Harman curve, yes? And no worry about distortion, yes?

Well me, I was very underwhelmed by it. This was at CanJam NYC 2022, in their listening room, and adequately powered, and there was no eq'ing going on.

I went back and forth between that and my favorite closed a few times, my Ether C Flow that I brought to the show, and their own Ether C Flow (both with the v1.1 tuning kit, and both sounded extremely close to each other; but also oone is going to say that tracks the curve). And my preference by a wide margin was confirmed.

I remain a Harman curve anti-fan.

 

[RHO]

Isn't one of Amir's headphone tests at 115 dB?

THD measurements are done up to 114dB. This has nothing to do with lack of perceived dynamics at lower listening levels compared to live concert levels.

 

[Feelas]

Why does anyone have to tell us what we should like, can’t we decide for ourselves. When I listen to different headphones that are designed to the Harman curve I feel as if they are flat and dull. I have attended hundreds of live performances in my life and they never sound anything like a headphone tuned to the Harman curve, voices are lively, horns and percussion pop, the music is fun. But we each get to decide for ourselves what we like, I say stop arguing and just enjoy the music…

Treat the Harman curve as an idealized point of reference which tells you how to read the graphs & where your best sounding graphs diverge to understand what kind of difference you like, exactly. No more second-guessing...

Why are people afraid of reference points?

 

[markanini]

I have not bought lesser headphones that track close to the target. But, I believe the DCA Stealth is considered to have excellent tracking to the Harman curve, yes? And no worry about distortion, yes?

Well me, I was very underwhelmed by it. This was at CanJam NYC 2022, in their listening room, and adequately powered, and there was no eq'ing going on.

I went back and forth between that and my favorite closed a few times, my Ether C Flow that I brought to the show, and their own Ether C Flow (both with the v1.1 tuning kit, and both sounded extremely close to each other; but also oone is going to say that tracks the curve). And my preference by a wide margin was confirmed.

I remain a Harman curve anti-fan.

I didn't say it was impossible, In fact I posted pie chart from Harman representing the 36% of people who find their target to be dissatisfactory. So far you're the first one sharing a relevant anecdote, following my call. Extended listening beyond a demo would be better. Ideally with a 105Hz low shelf to adjust bass levels to your individual preference. But I can definitely take your story at face value. I think hardly anyone generally in favor of the Harman target would have a problem with you.

BTW many will view statements like "I remain a Harman curve anti-fan" and "lesser headphones" as childish If you care come off as reasonable.