Good somewhat neutral mixing headphones with good bass extension for max. 350€?

[FireEmblem]

This was tested by Harman and especially the bass and upper midrange/treble showed quite a large dB range for personal preference.
Add to that seal issues and you have a mix.

Ah, I didn't mean preference but objective measurements.

Like, did they ever compare individual ear measurements with test fixture measurements to determine how close or far (= usable) the test fixture measurements were?

 

[FTB]

harman = good, crin = bad

that's basically your take. if you actually looked at his target and compared it to harman you wouldnt be saying anything like that. You're just negatively biased towards him and positively biased towards harman. Im not saying harman is bad, but even if the target is not 100% to your preferences, you would be blinded into liking it due to your subjective bias, which is why blind testing is so crucial.

I don't think there is much to gain with an argument here. I've laid out all my points here, if you want you can read them, if you agree then nice, if you disagree then that's nice as well let's agree to disagree.

I used to use Crinacle graphs in the past, before he made his targets.
It is said here, his targets like IEF blabla are based on his only perception, not research and saying B&k 5128 is best is not proven : https://crinacle.com/2025/02/05/the-new-2025-ief-target/

One people benchmark versus several people benchmark, I know what is scientific or not.

Curves made by reviewers are totally subjective and based on one people opinion.

 

[FTB]

Ah, I didn't mean preference but objective measurements.

Like, did they ever compare individual ear measurements with test fixture measurements to determine how close or far (= usable) the test fixture measurements were?

You can see this from page 29 to 38 : https://danishsoundcluster.dk/wp-content/uploads/2022/11/Olive_DSD_2022.pdf

 

[solderdude]

The targets used with measurements is preference based. An averaged preference.
So any EQ based on measurements thus consists of interaction between the headphone and test fixture + target.
When one looks at the 'reference curve' of raw measurements or the 'compensated' measurements you thus are looking at measurements + preference (the target) and that's what EQ is based on + measurement circumstances.

Like, did they ever compare individual ear measurements with test fixture measurements to determine how close or far (= usable) the test fixture measurements were?

Yes, there is tons of measurements like this showing individual differences (EDRP measurements on real heads) that can differ 10 to 20dB.
This, however, is not preference but measurements only at the ear drum.
They differ from test fixtures (even 5128) and on top of that there is human preference (the target part of the correction curve).
Very extensive research has been done by measurement fixture companies, universities and companies like Harman.
Most of it is behind pay walls and very technical.

 

[FireEmblem]

You can see this from page 29 to 38 : https://danishsoundcluster.dk/wp-content/uploads/2022/11/Olive_DSD_2022.pdf

Yes, there is tons of measurements like this showing individual differences (EDRP measurements on real heads) that can differ 10 to 20dB.
This, however, is not preference but measurements only at the ear drum.

Page 33 states that the largest deviation in open back headphones is only 1.8dB

But the graph on page 36 where fixture is compared with measurements from the humans, differences seem to be bigger, sometimes 10dB.

Soo when I see something like this on RTINGS:

Does "averaged and compensated" mean they already considered this so that one can expect that the deviations are smaller? Because if it's only 3-4dB that's acceptable I think.

By the way, it's crazy how much Sonarworks does improve my Superlux HD681, while the changes for HD650 are super subtle.

HD681 sound much more balanced with it. @kemmler3D recommended to prioritize balance, which now makes sense to me after hearing it in action even though absolute measurements are unknown. Sonarworks state on their website that because of the +/-5dB variations in the bass and treble in Superlux HD681 units, they'd recommend to get it measured, but of course nobody will pay 150€ measurement for a 20€ headphone where you can't even change cables if something breaks.

Buuut, if we assume that the said EQ curves from above are averaged and may only have deviastions of +/- 3dB per person if used as reference, and that it's not too difficult to review bass and treble by ear in terms of whether or not you got a unit that is more excessive or tamed in that region, shouldn't be Superlux HD681 + EQ a worthwhile combination still?

I think errors could be pinned down to maybe +/- 3dB. I could turn that 20€ headphone really into something great with fantastic lowend. And I like Kemmlers idea of just prioritizing balance rater than absolute measurements. The only advantage of the HD650 I'd see then is the wider soundstage.

HD650 have the greater soundstage, but Superlux let's me hear instruments more close to me which is good for noticing and mixing details. And since I will probably stick with the Adam T7V, they cover the wide stereo imaging thing. It's crazy because I for the first time ever witnessed the phantom center image today with the T7V. It was as if a violin was playing right in front of me in the air between both speakers. Didn't get that with the T5V due to the smaller drivers perhaps, but I'll play around with positioning and see if I can get the same effect.

(Edit: got the phantom center image working with T5V too, just needed to optimize the placement)

BTW what's funny, even though totally offtopic, is that T5V reaches just as low as T7V, although official data sheet says 45Hz. But both end exactly at 39Hz (sine wave generator).

 

[kemmler3D]

Page 33 states that the largest deviation in open back headphones is only 1.8dB

That's *standard* deviation which is a statistical measure which tells you how spread out the values are. 1 standard deviation tells you that ~68% of the values are within that range plus or minus, but it doesn't tell you what the maximum / minimum differences are.

 

[FireEmblem]

What's up with measuring the ear by test-tones? The sheet showed that Harman has an app called "HARMAN PERSONI-FI", and my JBL in ears use something similar.

It would play different tones in different frequencies and you can then let it know when you can't almost hear it anymore, and according to the results the headphones are adjusted to your individual ear. (JBL doesn't even do that, the app just claims that it measures your ear and then adjusts stuff).

Is that somehow reliable or just a nice gimmick?

I mean if having a balanced output is the goal, wouldn't that be a great method?

 

[solderdude]

Test tones with headphones is very unreliable above a few kHz.
Test tones for low frequencies are also unreliable for many reasons.
If one would need to measure the 'response' it would need to be done with narrow band noise.

HD650 needs a little correction.
HD681 needs a LOT of correction and there are several different HD681 which makes it more difficult.

 

[FireEmblem]

HD681 needs a LOT of correction and there are several different HD681 which makes it more difficult.

Hmm. Sonarworks rates HD681 at 6.5 without and 6.9 with average reference correction. 8.4 with individual measurement.

HD650 is rated 8.2 without average reference correction and 8.8 with.

But I feel the jump from uncorrected to uncorrected with HD681 is way more than just +0.4 It's night and day. Treble gets tamed, everything becomes more clear and balanced, upper mids come out better as a whole rathe than punctual. They sound much closer to the HD650 with EQ.

 

[solderdude]

Those numbers are nonsense.
Ignore them.

 

[FireEmblem]

HD681 needs a LOT of correction and there are several different HD681 which makes it more difficult.

RTINGs has updated the sound tests for Superlux HD681 in december 2024 and I've got a new pair because my old broke after many years.

So it should be pretty save to use their measurements, shouldn't it? They also offer EQ settings right away, that's pretty cool.

And according to Sonarworks, who measured different units of it, the +5dB deviations only appear in the sub bass and after 10.000Hz,
shouldn't it be possible to tell by ear if it sounds balanced or make minor adjustments in these regions?

I assume the spots themselves stay in the same place, only the amounts vary between the units?

 

[solderdude]

When you like what you hear then yes use it.
The new HD681 differs from the old one. What EQ cannot fix are resonances and the high distortion at 4kHz.
Then there is the HD681AIR as well.

 

[FireEmblem]

When you like what you hear then yes use it.
The new HD681 differs from the old one.What EQ cannot fix are resonances and the high distortion at 4kHz.
Then there is the HD681AIR as well.

I'm reading into how RTINGs is measuring things, and they test each headphone pair with 5 different people and measure in their ear to get a standard deviation score. So according to them everything 7.5 or higher is in the green best area.

Superlux HD681 got a score of 7.8, so apparently the deviations between people are not too big with this model which means the measured frequency response graph could be used as a good reference.

HD650 is a bit lower (7.3), but perhaps makes largely up for it by having more consistent units.

 

[olds1959special]

The only headphones I’d trust for actual mixing are Sony MDR-7506’s.

 

[solderdude]

The only headphones I’d trust for actual mixing are Sony MDR-7506’s.

They are not. These are monitoring headphones and not very comfortable either.
But here too... if one is fully familiar with them and knows how they 'translate' they could be used for that purpose (or at least combined with monitors).
The fact they are often found in studios is because they are very sturdy, easy to drive, decent isolation, above all CHEAP and have good mids (important).
For the same reason one can also find many DT770 in studios but these are less 'accurate'.

I'm reading into how RTINGs is measuring things, and they test each headphone pair with 5 different people and measure in their ear to get a standard deviation score. So according to them everything 7.5 or higher is in the green best area.

Superlux HD681 got a score of 7.8, so apparently the deviations between people are not too big with this model which means the measured frequency response graph could be used as a good reference.

HD650 is a bit lower (7.3), but perhaps makes largely up for it by having more consistent units.

Stop looking at all the numbers.
Open headphones (HD650) as well as semi-open (HD681) score good on measured consistency on heads (not at EDRP).
Closed one do not fair that well unless they have ANC engaged.

Measured consistency (at EDRP, ERP or EEP) is not the same as how sound is perceived (preference).

 

[olds1959special]

They are not. These are monitoring headphones and not very comfortable either.
But here too... if one is fully familiar with them and knows how they 'translate' they could be used for that purpose (or at least combined with monitors).
The fact they are often found in studios is because they are very sturdy, easy to drive, above all CHEAP and have good mids (important).


Stop looking at all the numbers.
Open headphones (HD650) as well as semi-open (HD681) score good on measured consistency on heads (not at EDRP).
Closed one do not fair that well unless they have ANC engaged.

Yea, I know how they translate, because I’ve used them for years for monitoring. I use HD650’s for reviewing mixes now, they are more comfortable.

 

[FireEmblem]

Stop looking at all the numbers.
Open headphones (HD650) as well as semi-open (HD681) score good on measured consistency on heads (not at EDRP).
Closed one do not fair that well unless they have ANC engaged.

I was reading into this because earlier in the thread it was mentioned that headphone measurements wouldn't be so reliable due to the fixture measurements deviating from human heads. And since I asked how big these differences are, I'm happy I found out where to look for it so that I have some certainty.

Is there any free, system-wide EQ plugin that's recommend as the best one around here?

 

[FTB]

Rtings don't follow Harman research when measuring headphones, if it is what is important to you.

 

[FireEmblem]

Rtings don't follow Harman research when measuring headphones, if it is what is important to you.

I want them to be as flat as possible, so I'd just even out everything that the measurements show above or below midline, right?

If the headphone has measured +5dB from 100-200Hz then I'll do -5dB in that region?

 

[olds1959special]

For the same reason one can also find many DT770 in studios but these are less 'accurate'.

I use the DT770-M as well because of its good isolation and the cable is longer.