• 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!

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

Been reading through this read and trying to understand the topic of neutrality.

Considering the previously mentioned "Circle of Confusion" described by Floyd Toole, how can we use terms like objective/neutral/transparent? A particularly target curve like the Harman curve is a summation of subjective preference. Doesn't that render it intrinsically void of objectivity?

I'm not understanding how claims of neutrality or transparency are made with confidence given the "circle of confusion," which in my laymen interpretation seems to mean that neutrality and transparency aren't possible. I feel inclined to believe that all systems are ultimately "colored," rendering further coloration perfectly acceptable and transparency/neutrality meaningless. Maybe I should ask in another thread? I felt it was related to target curves, which are used in Amir's reviews to determine adequacy.
 
Been reading through this read and trying to understand the topic of neutrality.

Considering the previously mentioned "Circle of Confusion" described by Floyd Toole, how can we use terms like objective/neutral/transparent? A particularly target curve like the Harman curve is a summation of subjective preference. Doesn't that render it intrinsically void of objectivity?

I'm not understanding how claims of neutrality or transparency are made with confidence given the "circle of confusion," which in my laymen interpretation seems to mean that neutrality and transparency aren't possible. I feel inclined to believe that all systems are ultimately "colored," rendering further coloration perfectly acceptable and transparency/neutrality meaningless. Maybe I should ask in another thread? I felt it was related to target curves, which are used in Amir's reviews to determine adequacy.

Actually the "circle of confusion" relates to criteria that we use to assess gear for colorization, since records made with non-neutral gear are played on non-neutral hardware and being taken as a measure of correctness for further experiments. Entirely different thing, I'd say.

The way that measurements are done in regards to Harman it doesn't fit here - as such, on uncoloured material, the averaged curve is supposed to be statistically the closest as possible for everyone to expected FR. Perhaps naming Harman "the curve that most can buy and not be disgusted" would serve the purpose. FWIW nobody presented a better theory on that for now.

And anyway, headphone FR personalization is the only real way forward.
 
depending on one's mood...it's just like henta*, see you just start on the normal stuff and then you started getting into all kinds of weird tags and before long you realize vanilla is the best...the thing to takeaway here is that it doesn't matter as long as you like it
 
Actually the "circle of confusion" relates to criteria that we use to assess gear for colorization, since records made with non-neutral gear are played on non-neutral hardware and being taken as a measure of correctness for further experiments. Entirely different thing, I'd say.

The way that measurements are done in regards to Harman it doesn't fit here - as such, on uncoloured material, the averaged curve is supposed to be statistically the closest as possible for everyone to expected FR. Perhaps naming Harman "the curve that most can buy and not be disgusted" would serve the purpose. FWIW nobody presented a better theory on that for now.

And anyway, headphone FR personalization is the only real way forward.

I think I see what you're saying. They are two entirely different things. I was conflating points unnecessarily.

What I'm having trouble understanding is when review's call a particular headphone or speaker's tonality "correct" or "incorrect," "science" based, transparent, neutral, etc., and use adherence to a curve or lack thereof to call something "objectively good" or "objectively bad" when, as you said, it is based on subjective expected FR with some particular conditions.

I went on a tangent to further express my trouble understanding objectivity here given the existence of the circle of confusion.

I think the idea of headphone FR personalization is really cool actually. But that would throw objectivity out the window too, wouldn't it?
 
What I'm having trouble understanding is when review's call a particular headphone or speaker's tonality "correct" or "incorrect," "science" based, transparent, neutral, etc., and use adherence to a curve or lack thereof to call something "objectively good" or "objectively bad" when, as you said, it is based on subjective expected FR with some particular conditions.

While I'm not a fan of scoring headphones against a particular target according to a mathematical equation, given the typical sort of variation you see across dummy head / test rig measurements and across listeners, whichever target you prefer (DF, Harman, Crinacle, Oratory's "Optimum Hifi", whatever), they all share a basic understanding of how the human ear works and are all variations around what is anatomically sensible.

When I see this : https://headphonetestlab.co.uk/test-results-manufacturers-a-d-bw-px7, https://www.rtings.com/headphones/1-4/graph#1619/4012
It's quite obvious that this was designed for aliens as the human ear canal resonance just doesn't work like that. It's very objectively rather terrible. Dummy head measurements are nowhere near inaccurate enough relative to what would happen on someone's head that the pair above would suddenly turn from a pumpkin into Cinderella's coach.

Besides, Harman's research is predictive of user preference, at least to a degree. It's likely that its predictive value fails once we deal with decently engineered headphones which are variations around "good enough" (HD560S, HD650), or headphones with inconsistent FR across listeners for various reasons (seal, pad compression, etc.), but in the current context when headphones such as the above exist, it's very relevant IMO.

I think the idea of headphone FR personalization is really cool actually. But that would throw objectivity out the window too, wouldn't it?

I don't know what is meant by "FR personalisation", but if it is about adjusting the FR output according to a user's own anatomy, then it might actually turn out to be the very most objective thing to do.
 
I think the idea of headphone FR personalization is really cool actually. But that would throw objectivity out the window too, wouldn't it?

Not quite, since in theory it is possible to measure ear characteristics and provide EQ that results in "flat" perceived response. It'd only depends on what you choose to target when personalizing - not unlike how people are tuning the house curves with speakers.

Also it is necessary to discern massive tonal faults and light discrepancy from perceived flatness - music ain't no magic and missing a bassline due to a botched FR might completely changed the perceived emotionality of a music piece.
 
Dummy head measurements are nowhere near inaccurate enough relative to what would happen on someone's head that the pair above would suddenly turn from a pumpkin into Cinderella's coach.

This is the thing that I have trouble with. When taking into consideration the immense number of people that consider some pumpkins to be Cinderella's coach and vice versa, it seems to me you can say that dummy head measurements can accurately/consistently/etc measure frequency response and correlate to perception. What you can't do based on those measurements is "objectively" make a statement on whether it is a pumpkin or Cinderella's coach. So, if I understand you correctly, I can see how a target curve is useful if you want to appeal to the statistical masses. Am I right to say what I'm saying though, which is that this appeal is in no way actually true objectivity or anything like that?

Not quite, since in theory it is possible to measure ear characteristics and provide EQ that results in "flat" perceived response.

This doesn't make sense to me either, unless a speaker or headphone's frequency response is literally a flat line? How do you determine what is flat? I've seen some attempts at explaining this and they all have converged on "it doesn't actually exist."
 
This doesn't make sense to me either, unless a speaker or headphone's frequency response is literally a flat line? How do you determine what is flat? I've seen some attempts at explaining this and they all have converged on "it doesn't actually exist."

And here the two get mixed up: while you can't necessarily make recordings sound "flat" (since their flatness is lost during mastering & mixing and cannot be recovered), you can - at least both AKG and Nuraloop argue - measure one's personal ear anatomy and shape the curve so that EQ + ears EQ end up in a flat perceived signal. The closest we can get is to hear the exact flaws that were put during record cutting and no more.
 
This is the thing that I have trouble with. When taking into consideration the immense number of people that consider some pumpkins to be Cinderella's coach and vice versa, it seems to me you can say that dummy head measurements can accurately/consistently/etc measure frequency response and correlate to perception. What you can't do based on those measurements is "objectively" make a statement on whether it is a pumpkin or Cinderella's coach. So, if I understand you correctly, I can see how a target curve is useful if you want to appeal to the statistical masses. Am I right to say what I'm saying though, which is that this appeal is in no way actually true objectivity or anything like that?

People's evaluation of a pair of HPs' sound may have little to do with how they actually sound but external factors (ie they may see Cinderella's coach instead of a pumpkin just because it feels comfortable, is expensive, well built, whatever - basically external factors will bias their judgment of sound quality, something extensively researched with speakers I believe - Mr. Toole may have something to say about this ?), which is why Harman's research attempted to replicate headphones virtually on another pair to make it an effective blind test, and showed that their methodology had a decent match with the real headphones, at least to a degree (as I said, I'm not sure that Harman's research predictive value is superbly good when it comes to predicting whether someone will prefer a pair of HD560S over a pair of HD650, but it certainly would be if we were to compare the PX7 above with another pair of Harman tuned headphones).

What I'm saying though is that this is a moot point in the case of the PX7 : its response simply doesn't correspond to how the human ear works, regardless of the target you prefer. The ear canal gain region just flat out doesn't correspond to the sort of FR you see at people's drum reference point when actually measuring their own ears, and the deviation is so huge that there's probably not a single human on the planet for whom that would sound optimal in that range. Basically, both its general shape, and the degree of departure from various targets, are wrong.
 
Considering the previously mentioned "Circle of Confusion" described by Floyd Toole, how can we use terms like objective/neutral/transparent? A particularly target curve like the Harman curve is a summation of subjective preference. Doesn't that render it intrinsically void of objectivity?

Apparently, if you somehow manage to assemble enough people who think or perhaps perceive something in a similar, subjective manner, it becomes objectively useful in the grand scheme of things, especially if the outcomes suit your narrative or ultimate commercial interests.

Otherwise, it's absolutely useless data and should be disposed of post-haste...
 
Word!

But some use it to approve or disaprove the perfomance of a headphone and call this judgement "based on science", while it is actaully only based on once taste.
This is getting boring.
 
I think a reference curve is very useful, you can look at deviations from it and get some idea what to expect a headphone to sound like. I don't like the use of it as a "target" however, as the implication is that deviation from the curve is a fault, and that all manufacturers should be trying to perfect Harman compliance or face bad reviews.

I can't think of anything more boring for this hobby than all headphones sounding the same.

Where I would like to see headphones measured and improved across the board is distortion, I don't think there is any argument that distortion can be a feature, where as a unique frequency response is exactly that in my mind.
 
I can't think of anything more boring for this hobby than all headphones sounding the same.
Why should headphones be excused from accurate reproduction? In an ideal world all headphones would sound the same as there can only be one 'correct' frequency response. Whether we can actually achieve that response in practise, or even measure it accurately, is another matter ...
 
Whether we can actually achieve that response in practise, or even measure it accurately, is another matter ...
We can't (yet?) because we don't know what it is.
So it's up to subjective preference. We have rough estimation of what that preference curve looks like for many people. But it is far from an absolute fixed representation of "accurate sound".
So, as long as there quite a bit of subjectivity involved, I think it is quite OK to give headphones their own characteristics. None of them are wrong as long as the deviation from what most people find acceptable/pleasant sound isn't too big.
 
The thing I find odd is that, for their initial research, Harman used the superb Revel F208 speakers in a great room.

The curve this produced was the basis for listening tests, with listers asked turn up/down bass and treble.

As is clear from Amir’s review of these speakers,there’s a drop off in sub-bass with these speakers.

https://audiosciencereview.com/forum/index.php?threads/revel-f208-tower-speaker-review.13192/

I don’t understand why they didn’t integrate a sub into the ‘great room’, EQd to be flat from where the sub-bass starts to roll off.

Then ask the listeners if they liked that.
 
Back
Top Bottom