[ADUHF]

I've had trouble figuring out how close cans are to F-M. A lot of the measurement sites apply compensation curves, and it's hard to figure out what the true measurement it.

I definitely hear you on that, bigshot.

It would make things a little easier if everyone used the same type of compensation curve for their measurement rigs. Ie, diffuse-field, free-field, etc.. One of Harman's goals though was the creation of a new compensation curve that better reflects the subjectively (and also objectively, if you accept the "room sound" theory) preferred tonal qualities of actual listeners.

One of the challenges though with the Harman curve is that there isn't a really straightforward way for the manufacturers of other measurement systems (than the one used by Harman) to reproduce it. In part, because it is based on both subjective and objective measurements. And also because there are different preferences for different demographics.

The methods for creating diffuse and free-field curves is pretty straightforward. You put the rig in an anechoic chamber. Take a series of DRP measurements from a tone generator at a specific distance, at different frequencies and different angles. And after a little computer processing to collate everything, you've got a beautiful new set of curves tailored specifically for your rig. Eezy-peezy.

You can think of the Harman target as an "offset" of the diffuse or free field curves, if you like. And try to come up with a rough approximation for different rigs that way. But there will probably always be some degree of error with a simple approach like that. Because it's not really modeling all the subtle nuances of sound in an actual room.

Another method might be to create a reference headphone, which is electronically calibrated to precisely match the Harman curve (on Harman's measurement system). Which you could then plop onto any measurement rig of your choice. And just take some raw DRP measurements from it to derive a new curve for that particular rig. I see some potential complications with that approach as well though.

What I'd really like to see are raw DRP measurements of actual rooms with good loudspeakers. ("Good" means anechoically flat to me.) Once we see what the actual room data looks like, then we can figure out what, if any, offsets may be necessary for subjective preferences after that. But I think the room measurements are where we really need to begin.

The best approach I've been able to come up with in the near term is sort of a variation on the reference headphone approach. Instead of using a single reference headphone though, what I try to do is infer what a subjectively-preferred room response might look like for a particular rig, based on the raw DRP measurements of a collection of different headphones which roughly approximate a "room sound".

Imo, the Harman curve is like a big sign-post, which is pointing us in the right direction... But it is not the final destination. And more work still needs to be done (esp. on the measurement side) to get us closer to that eventual endpoint.

 

[bigshot]

The other thing to remember is the Harman Curve is averaged from a range of preferences, not a specific target. It's a bell curve that reflects "typical" tastes. You personally might be on one end or the other of the bell curve and prefer something else. Like all response targets, it's a good baseline place to start, then you add salt and pepper to taste.

My Oppo PM-1s hit my sweet spot pretty squarely. Just a 2dB tweak in the upper mids and it's perfect. Close enough to not need to bother with EQ. Of course the sub bass isn't the same as speakers, but no headphones can do that.

 

[castleofargh]

What I'd really like to see are raw DRP measurements of actual rooms with good loudspeakers. ("Good" means anechoically flat to me.) Once we see what the actual room data looks like, then we can figure out what, if any, offsets may be necessary for subjective preferences after that. But I think the room measurements are where we really need to begin.

That's assuming a stereo track on speakers in a room will be anything like listening to that track on headphones. But in practice the changes go well beyond frequency response.
IMO we have a lot of convincing data on speakers+room(thanks again to Harman). The problem we encounter isn't a lack of such data but the lack of transferable data for usual headphone stereo playback IMO. How the left channel doesn't reach the right ear, the subjective impact of not having room reverb when the track was mastered expecting some, how headphones cripple or bypass most of our HRTF for sounds at their virtual locations, how much of a subjective impact the lack of tactile bass or head movements have. Those are all the questions I believe we should be aware of. Looking for answers is IMO of limited interest because we can't make a race horse out of a donkey. Instead of trying to understand how to get the best "wrong" stereo, it is time to focus on how to get proper sound with headphones. FR is just a portion of the problem.
I have no idea if the future of headphone audio will remain like it is because audiophiles seem more retrograde than any other people invested in tech hobby. Or if headphones will become more and more the target of recording and mastering engineers, to the point where we will have "mastered for headphones" stuff, whatever that actually implies. Or if we'll simply go the road of doing speaker simulations on headphones and keep mastering albums so they sound best on speakers? I really don't know.
My vote goes for speaker simulation and figuring out the best(as in most practical for general population) method to integrate a given listener's HRTF into the process. If we can do that well, headphones would rapidly become the real hifi solution compared to speakers.

 

[ADUHF]

^ All good points, castleofargh.

And I agree that there is no way of capturing all the subtle nuances of sound in a room in a simple frequency response curve, like the Harman curve. And realize there are many who would like to leap-frog over that problem to more sophisticated solutions for recreating the sound of a room, or any other space for that matter, inside a pair of headphones. Those are all valid pursuits imo. And I have read about some of the developments along these lines on Inner Fidelity and other sites. As I mentioned above, the Harman curve is imo simply a big sign-post pointing in the right direction. But not the final destination.

The question of how to achieve better sound in a simple single-driver dynamic (or planar magnetic) headphone without alot of fancy DSP and HRTF tech is still an interesting and relevant one to me though. I'm interested in the other kinds of stuff you're talking about above. And may try to offer a bit of input where I can on that in the future. But I'm also (still) interested in the problem of finding a better target compensation curve (or curves) for assessing a single-driver headphone's tonal performance than the more antiquated diffuse and free-field curves. Because I don't think we've adequately addressed that question yet. And it's info I can use to help me find (and/or EQ) a better-sounding headphone today versus five, or ten, or twenty years down the road. If you want to describe that as "retrograde" thinking, so be it.

Imo, the Harman curve is a very big step towards that goal. But I think we can go further. Additional room measurements are an important piece of that puzzle imo. Because the DRP measurements that Tyll made in Harman's reference listening room were not comprehensive enough. And not done with speakers configured to an anechoically flat response.

There is some work that we (or I) can do without those measurements. But it would still be helpful to have the measurements for comparison.

 

[bigshot]

You've got fundamental principles you're trying to get right... signal to noise, low distortion, accurate timing, frequency response, and spatial cues... The first three are pretty much taken care of with modern digital technology. Response is the wild card with transducers, so we make an effort to address that. But the spatial aspect isn't even on the radar of most people, especially audiophiles. Home theater folks are way out in front of audiophiles there. I agree with Castle that the future of home audio probably lies in signal processing and synthesizing space with headphones. But I don't think that the way we mix and master will really shift entirely away from speakers though, because there are huge dangers in trusting any set of headphones when it comes to bass response. I've noticed how some headphones can suck up bass and sound fine, but when you play back on speakers, or even on a different pair of cans, the bass can explode into an overdriven mess. The problem with mixing with headphones isn't so much the space, although that is certainly an issue, it's more about striking consistent balances that work across the board with a variety of different playback situations. You can fine tune better with speakers.

 

[jagwap]

AKG K371 matches it almost exactly.

The K371 and K361 match very closely, except they do not have the low frequency shelf filter boost,as they are designed for pro users, not consumers. Pro users tend not to want a distended bass.

 

[ADUHF]

The K371 and K361 match very closely, except they do not have the low frequency shelf filter boost,as they are designed for pro users, not consumers. Pro users tend not to want a distended bass.

Not really disagreein, as some of the higher-end headphones used for tracking and mixing can get quite bright in the treble and flat in the bass. Per some of my remarks above though, I was curious to see what the average response of several popular closed-back studio headphones might look like. So I computed an average response curve for five of the most popular closed studio monitors (in the $100-$200 range) sold by places like Guitar Center and Sam Ash for pro audio use. And here is the net result...

Here's a list of the headphones with links to the five Rtings FR curves used for the sampling...

AKG K371 (Right Channel)
AUDIOTECHNICA M50X (Right Channel)
BEYERDYNAMIC DT-770 PRO (Right Channel)
SENNHEISER HD 280 PRO (Left Channel)
SONY MDR-7506 (Right Channel)

All of the above have a fairly noticeable bass boost. The main difference between the AKG K371 and the rest is that its bass boost is lower in frequency than the others. There was also a slight imbalance in the bass on the Rtings K371 plots. So the right channel used for the sampling was slightly higher in amplitude than the left channel.

The net result of the 5 samples actually has a bit of a smile to it imo, which is denoted by the fairly pronounced peak in the treble at ~9 kHz. Suggesting that these may (as a group anyway) be somewhat better suited to lower volume use. Most of the headphones (except the Beyer DT-770, and to some extent the AT M50X) were also fairly rolled off though in the upper treble, above 9 kHz. Which is also interesting.

To keep this on topic, most of the above cans are somewhat close to the Harman curve imo. Though they all have some coloration (which is different for each HP). And are probably a bit on the brighter/smilier side of neutral as a group, esp. in the 9 kHz region.

 

[bigshot]

The only headphones I've ever seen in recording studios are beater cans used for isolation during tracking. They're usually $50 closed cans. I've never seen any better headphones in the studio. I think the term "pro" here is applied to impress non-professionals, not to indicate that professionals use them.

 

[ADUHF]

The only headphones I've ever seen in recording studios are beater cans used for isolation during tracking. They're usually $50 closed cans. I've never seen any better headphones in the studio. I think the term "pro" here is applied to impress non-professionals, not to indicate that professionals use them.

That may be so bigshot. I don't think Rtings has plots for the lower-cost cans you're referring to though. Unless you count the Tascam TH-02.

The K371 is still sort of new. But most musicians and pro audio users should be fairly well-acquainted with the headphones in the above sampling. Because they are some of the most popular studio monitors sold in the $100-200 range. I've edited my last post above to hopefully make that a bit clearer.

 

[ScareDe2]

I don't know if the HD250 linear II measure close to the HD250 linear 1, but it is quite close to Harman response:

I always felt the HD250 was the most neutral headphone I ever owned/auditioned.

 

[bigshot]

The K371 is still sort of new. But most musicians and pro audio users should be fairly well-acquainted with the headphones in the above sampling. Because they are some of the most popular studio monitors sold in the $100-200 range. I've edited my last post above to hopefully make that a bit clearer.

I think what I am trying to say is that when they specify something is "pro grade" or "studio monitors" they don't mean that professionals use them in studios. They are higher end products for the consumer audiophile market. Professionals don't really use headphones as studio monitors. They use speakers. Home enthusiasts use headphones because they can't afford or don't have space for a proper speaker setup to do mixing and mastering.

 

[castleofargh]

^ All good points, castleofargh.

And I agree that there is no way of capturing all the subtle nuances of sound in a room in a simple frequency response curve, like the Harman curve. And realize there are many who would like to leap-frog over that problem to more sophisticated solutions for recreating the sound of a room, or any other space for that matter, inside a pair of headphones. Those are all valid pursuits imo. And I have read about some of the developments along these lines on Inner Fidelity and other sites. As I mentioned above, the Harman curve is imo simply a big sign-post pointing in the right direction. But not the final destination.

The question of how to achieve better sound in a simple single-driver dynamic (or planar magnetic) headphone without alot of fancy DSP and HRTF tech is still an interesting and relevant one to me though. I'm interested in the other kinds of stuff you're talking about above. And may try to offer a bit of input where I can on that in the future. But I'm also (still) interested in the problem of finding a better target compensation curve (or curves) for assessing a single-driver headphone's tonal performance than the more antiquated diffuse and free-field curves. Because I don't think we've adequately addressed that question yet. And it's info I can use to help me find (and/or EQ) a better-sounding headphone today versus five, or ten, or twenty years down the road. If you want to describe that as "retrograde" thinking, so be it.

Imo, the Harman curve is a very big step towards that goal. But I think we can go further. Additional room measurements are an important piece of that puzzle imo. Because the DRP measurements that Tyll made in Harman's reference listening room were not comprehensive enough. And not done with speakers configured to an anechoically flat response.

There is some work that we (or I) can do without those measurements. But it would still be helpful to have the measurements for comparison.

To be clear, I'm not trying to undermine the Harman target for headphones. It came out of the most comprehensive and serious research on headphone preferences I've seen. In the present context, I do believe that they made the right choices and asked the right questions. I especially liked the paper on non linear distortions and what we kind of already knew about THD being a really poor indicator of subjective preference at the typical levels found in OK headphones. I wish I could also try to measure non coherent distos from multitone and music samples, but even checking the paper by Temme and Brunet on the method to do it, I remain completely lost(too much math, not enough .exe for idiots like me).

I also think that the frequency response needs to be taken seriously for preferences, and saw no reason to doubt that Harman has come really close to what will remain the average preference under current circumstances.

But considering how at least half of my favorite albums are from a period where headphones didn't even come to mind for the mastering engineers, and the rest still was made using speakers as main playback tool, I do wish for a change in the headphone world that goes beyond frequency response and showing nice disto figures. But this is probably not the topic for that

not sure what you would do with flat speakers in an anechoic chamber?

 

[jaakkopasanen]

People reading this thread are probably interested in headphone rankings based on Harman preference score: https://github.com/jaakkopasanen/AutoEq/blob/master/results/RANKING.md

 

[ScareDe2]

Does this list have anything to do with the Harman target or is a distinct study based on random preferences? I see the Utopia is positioned 12th, not a neutral headphone. I don't think there is many discontinued vintage headphones in that list either. How would a Sony CD3000 or a Sennheiser HD250 linear 1 score? We don't know.

Also, what do they mean by warm and bright? You can have an headphone that is not neutral but doesn't sound neither warm nor bright? That is strange.

 

[Davesrose]

That's assuming a stereo track on speakers in a room will be anything like listening to that track on headphones. But in practice the changes go well beyond frequency response.
IMO we have a lot of convincing data on speakers+room(thanks again to Harman). The problem we encounter isn't a lack of such data but the lack of transferable data for usual headphone stereo playback IMO. How the left channel doesn't reach the right ear, the subjective impact of not having room reverb when the track was mastered expecting some, how headphones cripple or bypass most of our HRTF for sounds at their virtual locations, how much of a subjective impact the lack of tactile bass or head movements have. Those are all the questions I believe we should be aware of. Looking for answers is IMO of limited interest because we can't make a race horse out of a donkey. Instead of trying to understand how to get the best "wrong" stereo, it is time to focus on how to get proper sound with headphones. FR is just a portion of the problem.
I have no idea if the future of headphone audio will remain like it is because audiophiles seem more retrograde than any other people invested in tech hobby. Or if headphones will become more and more the target of recording and mastering engineers, to the point where we will have "mastered for headphones" stuff, whatever that actually implies. Or if we'll simply go the road of doing speaker simulations on headphones and keep mastering albums so they sound best on speakers? I really don't know.
My vote goes for speaker simulation and figuring out the best(as in most practical for general population) method to integrate a given listener's HRTF into the process. If we can do that well, headphones would rapidly become the real hifi solution compared to speakers.

Note that I'm not looking up Harmon curve studies right now, but from what I remember, the curves for speakers are a lot more consistent since it's more realistic modeling of intended environment. Headphone studies are a lot more varied. We do know there are differences in perception and preferences due to age (some of it due to reduced frequency range range, but also differences in preference in music choice). Harmon has noted all this and tries to make averages. Way back when I was really active on this forum, I did notice there could be heated arguments about what is the best brand headphone for metal or electronica (from people roughly in the same age range). Given that the outer ear can vary quite a bit with people, and given that it funnels certain frequencies, I think those are ultimate factors about why no one brand/model headphone will sound "neutral" to everyone. To continue what was recently raised in the soundstage thread: perhaps the ultimate approach would be to take whatever headphone and then apply advanced DSPs (ones that do measure your actual ear anatomy and try to do virtual modeling of a 3D space).