Is it time for a new Harmon Curve survey?

[castleofargh]

I think I already mentioned that I have the AutoEQ filters for the HiFiMan HE1000SE for the Harmon curve installed in HQ Player and did try it. Which started me on this whole discussion. The Harmon curve has way too much bass. even after cutting the suggested boost in half. The reduced highs became dim and the prescribed elevation in the 1-2kHz range sounds muddy.

It could be that the curve isn't for you. As said repeatedly, it's completely expected that a good chunk of the population will not prefer it. In that case, you cannot turn the rest of the world into preferring what you prefer. No matter how many experiments you manage to kick-start, it's not happening.

But here, there is a very obvious second possibility. I have to come back to my main preoccupation on this website, bad testing leading to whatever false conclusions and people building ideas and certainties on top of those false results.
AutoEQ takes online graphs, makes a curve from the amplitudes on pictures(when FR graphs are better served with amplitude and phase if possible so we can really predict the resulting EQ precisely), or whatever data Jaakko could get his hands on, and applies some correction that's named Harman for convenience.
The process itself isn't the most accurate, the Harman curve used online pretty much always is an extrapolation also from a graph on some PDF(I don't know how many people got the data straight from Harman, maybe nobody?).
Nearly none of the online databases for headphones used/use the coupler and mic that Harman used for its research. Meaning there is a need for some correction, and because we're talking about different couplers, that correction for one headphone won't just work on another. There is no simple one fit all answer in that case, like measuring a headphone with both rigs to create a compensation. It's a more complex, more dynamic relationship.
Last, but not least, your headphone isn't the one measured for AutoEQ. Having the same model should ensure a close result, but it's completely common to get maybe up to 2dB here or there, and more at both ends of the audible spectrum.

And I already said it several times now, when it comes strictly to low frequency, Harman themselves warn that the biggest variation comes from placement and seal and that it seems to strongly drive the differences in preference, so there is no magic rule of liking the amount or cutting it in half, you should always adjust that yourself to fit your age, air cut, possible glasses, pad fit(how old they are if soft foam types)... The bass is up to you!
And of course, with planar and any large diaphragms, people tend to not want as much bass AFAIK(just my educated guess for that last one, I never read anything on that in research).



Anyway, try contacting some admins and participants, but if what you heard, happened to be close to the Harman curve, and you didn't like it for reasons other than all the stuff I brought up, you can bet that no matter how many thousands of people you get to try and come up with a new average curve, that curve will not please you either (and it will be close to Harman's). Being an outlier is the easiest thing in the world. As a tall, left-handed, blue-eyed, white guy, I'm already in a box with wayyyyyy less than 1% of the world population. It's that easy.
Of course, when constantly reading reviews from people who entirely ignore differences between humans and keep claiming that this headphone has this and that sound, we tend to get pulled into that fake reality, and the less we know about all that acoustic and psychoacoustic stuff, the easiest it is to put ourselves as the reference for how everything must sound to others.
My advice, if you find some guy whose made up curve and description of headphones seem to match your own feelings, follow that guy and that curve. Just don't fool yourself thinking he's better, and his curve is more accurate objectively for the world.

 

[bigshot]

I don’t know if anyone’s mentioned it yet, but there are two versions of the Harman Curve. The earlier one had less bass.

 

[gregorio]

Ha ha, take a chill pill XD

That’s your best response to all my points, “take a chill pill”, when I’m already chill?

G

 

[Killcomic]

That’s your best response to all my points, “take a chill pill”, when I’m already chill?

G

You're the one who wants to argue, that's the best you'll get from me. I couldn't care less what you think, you know?

 

[bagwell359]

Some good ideas on method, but, I'd like a number of sub sets of listeners, with matching results:

Say:

* pro musicians acoustic
* audiophiles w/ lots of live acoustic music experience

would be the two I'd want to see/hear

Stuff like:

* bass head
* people with little/no acoustic music experience

are ones I would not be interested in, except academically.

The trend in bass since the 1970's has been more, more, MORE. I find that the Harman (I think its 2017) to be very bass heavy (like +3-4 db) under 80 Hz, and probably 1.5-2 db from 80-150 Hz. I think the 2.5-3.5k also seems a but much as well, so I do adjust accordingly. Of course FR is only a part of the picture, but that's another topic.

 

[bigshot]

I don’t think that people with supposed “trained ears” would be able to do a better job. Preference for response in headphones has more to do with the shape of your ear canal than it does with experience with analyzing sound. If the subjects were able to define their preferences with specific frequencies’ numeric names, that would speed up the process, but I think a series of A/B choices could arrive at the same place. It would just take a little longer.

Everyone knows what real life sounds like, because that’s the world we hear every day. But the point of Harman is preference, not accuracy. Everyone- even your mom- has a preference. And measurements are a better way of determining accuracy.

 

[bagwell359]

I don’t think that people with supposed “trained ears” would be able to do a better job. Preference for response in headphones has more to do with the shape of your ear canal than it does with experience with analyzing sound.

Don't think ear canal analysis is going to be part of the data set anytime soon. People raised on pop studio recordings and air buds are going to randomly equal educated ears? Interesting, certainly seems a necessary belief for some, and the opposite for others. Have a testing methodology in mind? The image that comes to mind is folks that drink ripple/MD 20-20 vs say folks that get paid to opine on wine.

If the subjects were able to define their preferences with specific frequencies’ numeric names, that would speed up the process, but I think a series of A/B choices could arrive at the same place. It would just take a little longer.

Harman has changed over time. That seems to be a problem for your theory that a series of A/B choices would end up at the same place.

Everyone knows what real life sounds like, because that’s the world we hear every day.

You mean someone who has never heard say the BSO or equiv can be trusted to know how that should sound like? Let's say a given person has a heavy playlist of heavy metal and EDM? You just brought up the ear canal, so 'everyone' is indeed not the same, ipso facto, out of the gate.

But the point of Harman is preference, not accuracy. Everyone- even your mom- has a preference. And measurements are a better way of determining accuracy.

Some people think accuracy IS a preference, that is the sort of subset I am talking about.

To many Harman is a statement of accuracy. One of my many goals (low tier) is to make the same point. It's not accuracy, its the averaged taste of a non controlled group. The speaker/room for the Harman - Salon 2 - last I heard, in an average (read meh, untreated rectangular room) is the reference.

As I've written elsewhere Harman is a handy tool for starting my PEQ settings, however I never end up at those settings. Others have mentioned the same areas of disconnect with the suggested Harman curve for various cases that I have. Seems like a self selecting sub group.

 

[sendler2112]

It could be that the curve isn't for you. As said repeatedly, it's completely expected that a good chunk of the population will not prefer it. In that case, you cannot turn the rest of the world into preferring what you prefer. No matter how many experiments you manage to kick-start, it's not happening.

If a good chunk of the population will not prefer the Harmon curve as it currently is, ie majority, then this indicates that MOST PEOPLE will be happier with something else.

 

[castleofargh]

If a good chunk of the population will not prefer the Harmon curve as it currently is, ie majority, then this indicates that MOST PEOPLE will be happier with something else.

All the people who would prefer something else will most likely not agree on what that something else should be. But if there exist other fairly stable preference groups, it could be cool to have 2 or 3 other curves for 2 or 3 other major minority group preferences beside the Harman group. I'm for that, actually.

 

[GoldenSound]

The reason I started this discussion is because someone IS dictating the response curve to us. And I have read many reports by expert reviewers that are using programs that allow them to EQ that don't like the Harmon curve as it is

The thing here is that many people are misusing/misrepresenting what the harman research says.

The harman curve is NOT supposed to be a "This curve is great and anything that deviates from it is bad" answer. The fact that some do not like or do not prefer the harman curve itself is not an error, it's a big part of the research itself.

I'd strongly recommend reading through the segmentation paper and about how the harman research classified listeners into groups.

There are broad preference bounds/confidence intervals for which people will prefer different things. Some will prefer the harman curve but with elevated bass or treble, others the opposite. And the line should NOT be taken as a single target.

And then there's the issue of the fine grained data, such as exactly where to have the ear-gain peak or some fine grained shaping to the upper treble, which will be extremely listener specific due to differing HRTFs, and also due to non-HRTF related HpTF interactions.

Basically, the Harman research is great, and has given us a lot of incredibly useful information, but the people using it as a way to say "right, this exact line is how any good headphone should behave" are completely misunderstanding both the research itself, and the underlying ways in which headphones behave that mean it is not actually possible to have one single target or response curve that will satisfy all listeners, nor does the Harman research claim to be that.

 

[Daiyama]

Have fun discussing this with Amir.

 

[Resolve]

The harman curve is NOT supposed to be a "This curve is great and anything that deviates from it is bad" answer.

To add to this... anyone claiming this, or even suggesting that you should EQ exactly to this target is effectively doing it wrong. Yes, exactly those people who claim to care uniquely about objective information who hold this view are conveniently cherry picking small portions of the science while ignoring others because it fits their narrative and audio worldview. And, no matter how many times Dr. Olive himself needs to come out and tell them "it's just a guideline", they will still perpetuate that narrative.

The irony is that the pinna that's currently used with Harman 2018 isn't actually the one the research is based on, and so this really should have put an end to this perspective, and yet it seemingly persists. In my view, it's because of this extremely myopic view of the research that folks here (and in other places) will reject the Harman work, when really they should more appropriately apply that scrutiny to its misuse.

 

[colonelkernel8]

Some good ideas on method, but, I'd like a number of sub sets of listeners, with matching results:

Say:

* pro musicians acoustic
* audiophiles w/ lots of live acoustic music experience

would be the two I'd want to see/hear

Stuff like:

* bass head
* people with little/no acoustic music experience

are ones I would not be interested in, except academically.

The trend in bass since the 1970's has been more, more, MORE. I find that the Harman (I think its 2017) to be very bass heavy (like +3-4 db) under 80 Hz, and probably 1.5-2 db from 80-150 Hz. I think the 2.5-3.5k also seems a but much as well, so I do adjust accordingly. Of course FR is only a part of the picture, but that's another topic.

That's because bass is awesome. If you've ever attended a concert that wasn't an orchestra you may have noticed there's a LOT of bass you can feel in your chest. That's why the Harman curve is the way it is. If there was a new survey it would probably be even more juiced in the low frequencies.

 

[colonelkernel8]

Is it time? i want a double blind scientific analysis with at least 98.5% statistical significance.

and who eq anyway. straight line not good enough?

If you aren't using EQ (especially a modern DSP-based PEQ), then I'd say you can barely call yourself an audiophile. People who just buy extremely expensive equipment (or god forbid, cables) hoping that it sounds like they want is a credit card enthusiast, not a sound enthusiast.

 

[Whazzzup]

If you aren't using EQ (especially a modern DSP-based PEQ), then I'd say you can barely call yourself an audiophile. People who just buy extremely expensive equipment (or god forbid, cables) hoping that it sounds like they want is a credit card enthusiast, not a sound enthusiast.

lol