Neutral Circumaural Headphones

Technical Discussion
1

I’d like to start a couple of lists of neutral (or close to neutral) over-ear headphones. One for closed, and another for open. If you have any suggestions along these lines, please post them below.

I’m mainly interested in headphones which currently have measurements on the HBK 5128. And also on a GRAS system, for comparison to the Harman over-ear target. The headphones do not need to conform precisely to the Harman target. But they should be in the ballpark, and perceptually neutral (similar to in-ear response of a good pair of neutral speakers in a semi-reflective room) across most of the frequency spectrum.

The goal is to get a better idea of what a neutral response might look like on the HBK 5128 (and GRAS rig), by computing an average frequency response curve from several headphone measurements that are close to neutral. This would be used for comparison to other potential reference curves, including DF HRTFs, Harman’s over-ear target, the Headphones.com preference curve, and my own DF+SP model, among others. And as a potential cross-platform reference point.

This is not intended as a recommendation list.

3

I like to help friend but not sure if I can name headphone models that meet your criteria off the top of my head. If you aren’t aware of it already, there is an extensive repository of headphone measurements from various sources at AutoEq/measurements at master · jaakkopasanen/AutoEq · GitHub.

The oratory1990 directory is all GRAS. Those measurements are from oratory1990’s repository of EQ presets. The EQ preset files show the before measurement as well as give it a rating on how closely it matches the Harman Target. I think this type of rating is also available in the AutoEQ code repository, but I don’t know where.

And of course there is also a topic in this forum that contains measurements made by Resolve. Some are GRAS and some 5128. Your call what is considered neutral.

Hope that helps!

2 Likes
4

Indeed yes! Thanks for posting all the above links, AudioTool. Fwiw, I have also just added the following to Resolve’s measurements list…

This includes the Sennheiser HD 600, 650, 6XX, 660S, as well as the new HD 490 Pro, and closed HD 620S.

There are also more HBK 5128 measurements in Jude’s recent reviews and videos at Head-Fi, and in the most recent Sound Guys headphone reviews. And a fairly large number of GRAS measurements (useful for Harman comparisons) on ASR.

5

Both the AKG 371 and Modhouse Tungsten are tuned close to Harman.

2 Likes
6

Neutral for whom? There can be +/- 10 dB differences at various highly personal sets of frequencies between how one human ear canal modifies the sound of the same headphones vs. another human ear canal. The only point in looking for a single reference curve is to make mass manufacture of “reasonably good” headphones cheaper by applying a single design with no physical personalization. But it will not achieve neutrality for the vast majority of people, and should not claim to. Individual variation of +/- 10 dB perceived from the same headphone is not neutrality, it’s headphone roulette. :slight_smile:

2 Likes
7

Howdy again, AudioTool! :slight_smile: Fwiw, I think the AKG K371 and Mod House Tungsten are both pretty good suggestions, and definitely belong in this conversation. I have a couple questions I’d like to put to you and others though about these two.

I’m not that familiar with the Tungsten, and haven’t listened to them. And notice that they come with two different types of pads: hybrid and perforated. The perforated pads look closer to the Harman target in both Oratory’s and Resolve’s GRAS measurements. So that’s probably the one I would choose to include in the neutral open-backs. I’d like to hear some other opinions on this though. Especially from anyone who’s used the headphones with both sets of pads.

I’m more familiar with the K371, and have listened to them a number of times. And my impression is that they’re in the neutral ballpark, but they lack air and detail in the treble. And are a bit overdone in the sub-bass. The measurements are certainly open to interpretation. But I think you can also see this overemphasis in the sub-bass and lack of air in the treble on both Oratory and Resolve’s GRAS plots of the K371…

I’m not sure which version of Harman was used on Resolve’s graph above, because it looks a little whimpy in the bass. :slight_smile: But it’s worth notiing that the sub-bass on the K371 is higher than any other part of the headphone’s frequency response on this graph, including the peak in the upper mids around 3 kHz (the so-called “ear gain” region). And it noticeably overshoots the sub-bass on the Harman target on Oratory’s graph as well

I should add that I think the Harman over-ear target also lacks some air in the treble. So I’d probably be more inclined to include headphones in a neutral sampling that either meet or slightly exceed the Harman target in this area, rather than undershooting it to some degree, like the K371 appears to do.

I don’t think this is an especially controversial point of view. And it’s buttressed by the fact that many good headphones with a reputation for a neutral response also overshoot the Harman target in the upper treble or air region. (And also the fact that Dr. Sean Olive omits most of this upper treble/air region from his own Harman compensated plots and ratings.) But I know there are some folks who prefer headphones that are more mellow in this area than I do. So I’m open to hearing some other opinions on this.

I think I’d feel more comfortable including a headphone like the AKG K371 in a neutral sampling though if there was another closed-back that skewed a bit in the opposite direction, to help balance some of these features in the K371 out.

Please give me some of your thoughts on this if you have any.

I assume, btw, that most of the neutral open-back headphones will undershoot the Harman over-ear target in the sub-bass. And this is the main reason why I’ve seperated the neutral headphones into closed and open categories.

8

Welcome aboard abm0! And thank you for the link to the Griesinger video.

I am slightly familiar with some of Griesinger’s ideas on this subject. But have not had a chance to watch the video you referenced above yet. I will post a few of my thoughts on it and on your other questions above though once I have.

When it comes to questions of neutrality though, I mostly fall into the Floyd Toole, Tyll Hertsens, Paul Barton, Sean Olive, etc. camps in terms of my beliefs.

9

I’m not sure which version of Harman was used on Resolve’s graph above, because it looks a little whimpy in the bass.

Harman combined uses the bass of Harman 2013 + mids and treble from Harman 2018 so it has less of a bass shelf compared to Harman 2018

2 Likes
10

I have only briefly auditioned both the Modhouse Tungsten and AKG K371 so I can’t really comment further. Honestly though, individual HRTF starts to play a bigger factor the higher you go in frequency. And indeed Dr. Olive’s research shows that individual preference in bass level, if not perception, also varies. The very definition of what is a neutral headphone FR is something that is still being researched. Also there is only so much a headphone can do to be naturally tuned to any target. To match exactly requires PEQ. So a headphone that measures close to the Harman Target 2018, or Resolve’s Harman Combined, or the Headphones.com EQ target is about as good as we can currently do to objectively label a headphone as neutral.

1 Like
11

Neumann NDH 30 and Sennheiser HD 490 Pro might be worth looking at as they are designed for studio mixing.

1 Like
12

We may actually touch on exactly this topic in this week’s live stream. The question of “what is neutral” has been discussed by us in the past but perhaps not sufficiently addressed. I see a bunch of stuff here in this thread alone that indicates there’s still quite a bit of confusion on this, so we’ll try to hash some of that out shortly.

4 Likes
13

Welcome, Tara. And thank you for clarifying some of the different versions of the Harman curve above.

The blue curve on this graph is the Harman target I currently use for over-ear headphone measurements on the GRAS fixtures. This is from Sean Olive’s Twitter (now X) feed. And I believe it is the 2018 Harman target. (Feel free to correct me though if you think this is wrong.)

I think this is also the curve Oratory1990 uses on his GRAS PDFs, such as the AKG K371 graph linked above. Or something pretty close to it.

This appears to be the same curve used for comparison on Dr. Olive’s GRAS measurements. I think Dr. Olive has an HBK 5128 rig now as well. And I wish there was a list of his available measurements on both the GRAS and 5128 rigs.

1 Like
14

Thank you for your continued contributions here, AudioTool. I agree with alot in your last post. I see what appear to be some inconsistencies between the Harman curve and the current Headphones.com preference curve though. Enough at least that I think it could use some more work.

I don’t think the Harman curve is the best we can do either. But it’s (still) about all we’ve got to work with at the moment.

I also agree with Resolve that there is probably some confusion here and on other audio forums about what a neutral response is. But I think there’s actually a fair amount of consensus on this among the experts and luminaries in the field. And it essentially boils down to what I described in my first post about the in-ear response of neutral speakers in a semi-reflective room.

Paul Barton describes this as “room feel”. And the idea was also embraced by Tyll Hertsens of Inner Fidelity. And subsequently confirmed to a large extent by the Harman research*, which found that most people prefer a frequency response in headphones that is close to this type of sound.

This is consistent across most groups, though they may have somewhat different preferences for the amount of bass, midrange, or treble, depending on a variety of factors. And this is were alot of the research is still ongoing. But the basic principle of what is neutral is pretty much settled at this point. And it’s not something that 's likely to change much.

We may not have all the answers yet, but we have a pretty good idea of the direction we’re headed. And it’s mostly just about sifting out some of the finer details at this point.

(*This is why PSB and NAD headphones also hew fairly closely to the Harman target.)

15

NDH 30 and HD 490 Pro are also good suggestions. The NDH 30 is fairly dark in the upper mids and treble vs. Harman 2018, based on Ora’s graph. However, I think I’ve figured out a way of parsing some of the other close to Harman measurements so they can be included here, and will make a bit more sense.

I’ve gone through pretty much all of Oratory’s and Amir’s (at ASR) over-ear GRAS measurements btw, and identified a few dozen other neutral or near-neutral headphones to potentially be included and sorted here as well. And will try to post a more complete list of them soon.

16

The current Headphone Show livestream includes the topic of neutrality btw.

A nice soundtrack for neutral measurement browsing. :slight_smile:

1 Like
17

A great discussion around the definition of “neutral”, but after watching it my understanding hasn’t improved. It still sounds like something that is a subjective interpretation by the individual listener and the best objective definition is Harman 2018 or similar. @Mad_Economist even hinted at this when he said that most people define neutral as what they prefer, and Harman is essentially a preference curve based (at least initially) on in room FR of neutral speakers.

Unless I missed something, @Resolve? As I said, a great discussion but maybe I need an executive bullet point summary. :grimacing:

1 Like
18

Just to be clear, few would actually prefer the in-room baseline, same as flat DF. People like more bass generally, which is why the Harman curve is the way it is.

3 Likes
19

Agreed.

If I’m understanding Resolve correctly, then the dashed green curve on this graph is the baseline (in-ear) measurement he’s referring to, that people generally wouldn’t like because it favors the higher frequencies too much…


As Tyll Hertsen’s article on this explains, the green curve represents the in-ear response of a speaker (or speakers?) EQ-ed to a flat in-room response at the listening position. This was measured by Harman at the eardrum reference point (DRP) of a dummy head/mannikin they put together.

In a semi-reflective room, most good speakers normally have a sloped in-room response that ranges between about -0.7 to -1.0 dB per octave, above F0 in the bass. Or somewhere around -0.85 dB/oct on average. This is my estimate based on an average of 25 good loudspeakers with approximately flat direct responses. (Calculated value was actually -0.832 dB/oct, so I’ve rounded it off for simplicity.)

IN-ROOMSLOPE

The slope or tilt in the in-room response is due primarily to a speaker’s broader dispersion in the lower frequencies, which is reflected back to the listener’s position by the ceiling, walls, and floor of the room (see Floyd Toole’s video for more on this.)

By EQ-ing the speaker(s) to a flat in-room response at the listening position, Harman essentially removed the normal tilt that a speaker has in a typical, semi-reflective room (like in your home). You can see many more examples of this tilted in-room response in the spinorama graphs of good speakers here.

If you add the normal in-room tilt back in (which is something that can be done fairly easily with EAPO’s stacking and GEQ features), then you get something closer to the black curve on the first graph above…

SLOPEDINEAR

Harman’s version is a little more boosted in the lower/sub-bass and recessed in the upper-bass/lower-mids because the slope they used was their 3-knob preferred in-room loudspeaker response curve at the time, which was based on the movement of bass and treble shelves, as Dr. Olive explains here…

1 Like
20

Most everything about humans involves preference and ability variations – this typically results in bell curve normal distributions when graphed. One can often easily predict what will be perceived as normal for most people (e.g., Harman) and the upper/lower bounds for almost everyone, but not predict what a particular person will prefer. Individual variation follows from diet, environment, age, hearing damage, genetics, and more.

The electrical/measurements absolutist approach is bass ackward, as reports of “neutral” follow from the perception of just-noticeable differences. Music that sounds “the same” or “good” or “correct” to a specific person occurs because that specific person cannot distinguish between live music, a reference playback system, and the output of a given playback device. Given the heavy pre-and-post processing baked into non-acoustic genres, there often is no ground truth and there is no neutral.

If all you had was old-school vinyl LP records, then the technology limited the bass to prevent skipping. In the post-LP era people push frequencies up or down wherever they want. Even when they enjoy traditional bongo or taiko drumming and massive bass. Even when they enjoy flutes, trumpets, and piercing tones.

4 Likes
21

I apologize for not responding to some of the posts here in awhile. I’m still looking at graphs though, and trying to decide on the best way to break em down. And I’ve also deleted the list I had in the 2nd post, because it can no longer be edited and updated.

One list will be GRAS measurements close to the Harman 2018 over-ear target. Another list will be 5128 measurements close to the DF+SP model I use. There is likely to be some crossover between them. But I’ve already found a few interesting mismatches. And it will be interesting (to me anyway) to see what the 5128 DF+SP-tuned models look like measured on the GRAS. Just as it will be interesting to see how the Harman-tuned models measure on the 5128. Other lists could follow, but this is where I plan to start.

These are some of the headphones that appear close to Harman 2018 over-ear on the GRAS so far, along with a few of my notes. (This could change though.)

CLOSED & SEMI-OPEN HEADPHONES CLOSE TO HARMAN 2018:
Beyerdynamic Custom One Plus (velour pads, bass slider 3/4) - Response is a little bumpy, and peaky in places. But a pretty good match overall. There is a little peakiness in the upper mids around 2.5 kHz on Ora’s plot.
DCA E3 - Very good match to Harman 2018. Maybe slightly brighter in the low treble.
DCA Stealth - Slightly more bassy.
Focal Azurys - Slightly steeper overall slope.

OPEN HEADPHONES CLOSE TO HARMAN 2018:
Fiio FT3 (pleather pads) - A little warmer in upper bass / lower mids.
Focal Hadenys - A little withdrawn around 2 to 2.5k vs. Harman 2018.
HFM Sundara (2021) - Slightly withdrawn in the upper mids/low treble.
Onkyo A800 - Slightly withdrawn in some spots in the upper mids.
Sennheiser HD 600
Sennheiser HD 6XX
Sennheiser HE-1 - Withdrawn in upper mids.

Most of the open-back headphones are not a good match to Harman in the lower frequencies, so the comparisons are based more on the mids and treble on those.