Headphone EQ Suggestions

[solderdude]

I don't think bone conduction is an aspect for OE headphones. Is there any evidence it is. AFAIK for bone conduction the driver has to make direct contact with the skull (through thin layer of skin).

The biggest issue with IEM's is insertion depth, ear canal shape and depth as well as seal issues which is what makes them difficult to EQ.
This makes EQ for IEMs much more difficult than over-ear or on-ear.

 

[Kouioui]

Oratory's FAQ has this to say about the limitations of EQ:

Does this mean all headphones will sound the same after applying these EQ settings?
Yes.
Yes and no.
Actually mostly no.
Most headphones will sound much more similar than before, but I am explicitly saying that they will not necessarily be indistinguishable from each other - especially to a trained listener. It would be possible to make them sound much closer if I were to make measurements inside your actual ear. But then those measurements would not be applicable to anyone else other than you.

 

[solderdude]

Assuming none of the above-mentioned is a problem, it doesn't matter what software you use, as long as it does what you ask of it. You just have to equalize the headphones to the same frequency response, which you would need to confirm using a high-end measurement rig like the GRAS 45CA.

That's the thing I would like to know (a retorical question, but like to hear personal experiences).
Oratory measurements are all made so they measure flat, in other words tonally equally flat.
This would mean that IF the Grass 45CA coupler and the corrections that belong with it all different headphones with oratory (or Sonarworks or other headphone correction plots) would all have to sound similar in tonal balance.
So not the same as in other than tonal aspects but tonally.

Looking at the different EQ's one gets from 'headphone EQ software' they all cannot possibly be correct as in some ways they are similar but far from the same.
The only way to tell if Oratory (or Sonarworks etc.) are correct is if different headphones that are corrected using the same software all sound truly neutral and equal in tonal balance. If not then the measurements/correction cannot possibly correct.

Of course there is unit to unit variation and given the measurements I made with a few of the same HP's I would expect differences based on those measurement differences alone.

 

[solderdude]

Oratory's FAQ has this to say about the limitations of EQ:

Most headphones will sound much more similar than before

That is my experience as well (not using digital EQ)

The next question would be. Are differences between tonal balance between 2 or 3 headphones be smaller using Oratory, Sonarworks or other 'profiles'.

 

[thewas]

Assuming none of the above-mentioned is a problem, it doesn't matter what software you use, as long as it does what you ask of it. You just have to equalize the headphones to the same frequency response, which you would need to confirm using a high-end measurement rig like the GRAS 45CA.

Even if you do that on a high-end rig (Oratory1990 uses also a good one) this doesn't mean that they will sound the same for everyone, as everyone's ears have different geometry and volume which impacts on the sound arriving at the ear drum.

 

[flipflop]

That's the thing I would like to know (a retorical question, but like to hear personal experiences).

The personal experiences are documented in Sean Olive's AES paper 'A Virtual Headphone Listening Test Methodology', which was essential in the development of the Harman target curve.

So not the same as in other than tonal aspects but tonally.

An accurate frequency response is also important for a correct spatial reproduction of the sound. See for example 'Effect of Headphone Equalization on Auditory Distance Perception'.

Looking at the different EQ's one gets from 'headphone EQ software' they all cannot possibly be correct as in some ways they are similar but far from the same.
The only way to tell if Oratory (or Sonarworks etc.) are correct is if different headphones that are corrected using the same software all sound truly neutral and equal in tonal balance. If not then the measurements/correction cannot possibly correct.

You do know that oratory doesn't provide any software for his EQ, right? He only offers a list of filters, which can then be used with any PEQ of choice. Sonarworks don't show what filters they use, but they do show the compensation curves, which can be closely approximated with other EQ software such as Equalizer APO.

Of course there is unit to unit variation and given the measurements I made with a few of the same HP's I would expect differences based on those measurement differences alone.

I think this is the core of the problem, and it's the reason Sonarworks measure 5 headphones of each model and create EQ profiles based on an average of those 5 measurements. One can't expect to get a flawless EQ preset with the information contained in just a single measurement like oratory is offering. Lackluster QC is often too much of a problem.

 

[flipflop]

Even if you do that on a high-end rig (Oratory1990 uses also a good one) this doesn't mean that they will sound the same for everyone, as everyone's ears have different geometry and volume which impacts on the sound arriving at the ear drum.

Fair point, but headphones can also be measured at the DRP of humans. Considering how small the differences in HRTFs are, I don't think it would be wrong to say that a pair of headphones (or two pairs EQed to the same target with a HATS) would sound "(almost) exactly the same" to anyone.

 

[solderdude]

The personal experiences are documented in Sean Olive's AES paper 'A Virtual Headphone Listening Test Methodology', which was essential in the development of the Harman target curve.

Looking for personal experiences from members. I know OW research.

An accurate frequency response is also important for a correct spatial reproduction of the sound. See for example 'Effect of Headphone Equalization on Auditory Distance Perception'.

Indeed but beside the point here. The real question is what is accurate. And no I do not mean OW target curve but which EQ based on measurements is accurate. It can only be deemed accurate when headphones sound the same to most people.
Of course we all know this is not really possible.
What I want to know is when one has several headphones and EQ them to say Oratory, or Sonarworks or other 'correctional EQ' then how different does one find the tonal balance.
Are differences smaller for instance using oratory or Sonarworks with the same 'target curve' of course.

You do know that oratory doesn't provide any software for his EQ, right? He only offers a list of filters, which can then be used with any PEQ of choice. Sonarworks don't show what filters they use, but they do show the compensation curves, which can be closely approximated with other EQ software such as Equalizer APO.

Of course I know. One can used various software, for Sonarworks you need to use their software though.

I think this is the core of the problem, and it's the reason Sonarworks measure 5 headphones of each model and create EQ profiles based on an average of those 5 measurements. One can't expect to get a flawless EQ preset with the information contained in just a single measurement like oratory is offering. Lackluster QC is often too much of a problem.

I am willing to go further and state that depending on the measurement setup and used compensation you also get different measurements and different EQ.

The answer to Oratory's deviations (aside from people liking it or not) has already been given in post #22.
Yes, spread will also have an impact. Even in 3 HD650's I measured audible differences between 3 models even though these are some of the most constant quality headphones. The same I measured for Superlux models and X2HR models.
One of the open questions is how does Sonarworks do. I mean his test rig (but not compensation) probably doesn't differ much from mine.

Rtings now is migrating from their own target curve to Harman curve (and stops at 10kHz as does Oratory). Even the Harman target changed over the years so is still 'under development' it seems.

What works for your average teenager doesn't work for an experienced listener etc.

So basically I thought the one that has the least tonal balance between various headphones (so not make A sound like B) is probably closest to the best fitting curve for various headphones.

 

[Kouioui]

Looking for personal experiences from members.

Too many variables to say one measurement, profile, EQ implementation is best for everyone. I know from personal experience that what I hear on stage or in the studio performing music differs wildly listening from an audience perspective in a room.

If you're used to hearing everything close mic'd (or no mics) your brain tends to prefer that over sitting in a concert hall. Most recordings are meant to be played over speakers in a room while headphones put you much closer to the mics.

 

[thewas]

Fair point, but headphones can also be measured at the DRP of humans. Considering how small the differences in HRTFs are, I don't think it would be wrong to say that a pair of headphones (or two pairs EQed to the same target with a HATS) would sound "(almost) exactly the same" to anyone.

Only if measured at the ear drum or ear canal of the specific human, the transfer functions of human bodies and ears are quite different, a reason why binaural systems only work well with an individual ear calibration.
http://www.davidgriesinger.com/Binaural_hearing_and_headphones.ppt
Measured and EQed with a different ear they might sound more similar than before but there won't be a great match, especially at higher frequencies.

 

[Severian]

I think what I'd really like to experiment with is corrections to a target curve based on the HD650 response from the midbass on up. I'd be happy with that tonal balance on everything...

 

[solderdude]

My theory is that HRTF differs from person to person depending on many variables.
In the end if we have a band in front of us we all hear the same music. We don't hear it the same way but for each person a piano sounds like a piano and voice like a voice ... etc. Our HRTF is 'embedded' in the brain.

I know that we all don't and can't hear headphones the same way due to the many obvious reasons. That's not the info I am looking for though.

What I want to know is (and understand this differs from person to person and headphone to headphone) whether or not different headphones in one collection (that sound different) sound 'closer' to each other (and the target curve belonging to the method used) than when using the same headphones with a correction dreamed up by some other's 'compensation'.

So 1 individual that has played with various EQ profiles (and perhaps trial version of Sonarworks) with for instance Harman EQ have found headphones A, B and C to be closer to each other using Oratory, or Sonarworks (I use these 2 as example because they differ a lot in some cases) but EQ based on EARS (with various compensations) or Rtings or whatever they experimented with has first hand knowledge of which/who's compensation makes some headphones that individual owns sound closest to each other (not exactly the same)

There won't be many that have done such tests.

I know about the underlying challenges about sound reproduction, recording, near-far-diffusefield and various targets and why they exist and how they were derived.
That's not the info I seek.

First hand experiences of members that have experimented with more than 1 'headphone correction profile/program' and in which case they had to do the least 'tweaking' or found that after the profile was used they sounded the most similar.

I understand production spread, measurement issues and HRTF differing individually which also has an impact.

 

[Kouioui]

There won't be many that have done such tests.

I've tried quite a few and so far oratory's settings using APO/Peace come closest to what I hear sitting in front of my calibrated nearfield monitoring setup playing the same content. I trust mixes I make using my 6XXs to translate to speakers very well in overall tonal balance.

 

[flipflop]

Looking for personal experiences from members. I know OW research.

I struggle to see the value in sighted listening impressions that can't even be backed up by measurements.

What I want to know is when one has several headphones and EQ them to say Oratory, or Sonarworks or other 'correctional EQ' then how different does one find the tonal balance.

If you were to listen to the exact units measured and corrected by oratory, I think you would find them to sound extremely similar and very neutral as Olive's research would indicate. Are you suggesting this is wrong?

Are differences smaller for instance using oratory or Sonarworks with the same 'target curve' of course.

If you take a headphone and apply the same EQ to it with 2 different pieces of software, it should sound the same unless there's something wrong with the at least one of the programs used.
Maybe I'm not understanding your question correctly.

I am willing to go further and state that depending on the measurement setup and used compensation you also get different measurements and different EQ.

Yes, different measurement rigs will give different results. That's why it's a good idea to use the GRAS 45CA if the goal is to equalize headphones to the Harman curve.

One of the open questions is how does Sonarworks do. I mean his test rig (but not compensation) probably doesn't differ much from mine.

Sonarworks are notoriously secretive. The only thing I know is that they use a variation of the HTC, which seems to have too much energy in the bass and around 2 kHz.

Rtings now is migrating from their own target curve to Harman curve (and stops at 10kHz as does Oratory). Even the Harman target changed over the years so is still 'under development' it seems.

Right, but they HATS is different from the one used by Harman, so the curve has all sorts of quirks to compensate for that fact.
To be exact, their correction curve goes all the way up to 20 kHz, but they don't weigh frequencies higher than 9 kHz for their scores, because the measured response is unreliable past that point.

Even the Harman target changed over the years so is still 'under development' it seems.

Science is an ongoing process.

What works for your average teenager doesn't work for an experienced listener etc.

As demonstrated in the latest paper, both untrained and trained listeners preferred the Harman target over any of the virtualized headphones.

 

[thewas]

My theory is that HRTF differs from person to person depending on many variables.
In the end if we have a band in front of us we all hear the same music. We don't hear it the same way but for each person a piano sounds like a piano and voice like a voice ... etc. Our HRTF is 'embedded' in the brain.

The problem is that is part of this HRTF is bypassed when listening to headphones, thats why EQing them doesn't work as well as it does for loudspeakers.

What I want to know is (and understand this differs from person to person and headphone to headphone) whether or not different headphones in one collection (that sound different) sound 'closer' to each other (and the target curve belonging to the method used) than when using the same headphones with a correction dreamed up by some other's 'compensation'.

So 1 individual that has played with various EQ profiles (and perhaps trial version of Sonarworks) with for instance Harman EQ have found headphones A, B and C to be closer to each other using Oratory, or Sonarworks (I use these 2 as example because they differ a lot in some cases) but EQ based on EARS (with various compensations) or Rtings or whatever they experimented with has first hand knowledge of which/who's compensation makes some headphones that individual owns sound closest to each other (not exactly the same)

From my few personal experiences and from what I have read from similar experiences in various forums there is not one measurement rig that works as well for everyone, so for person A measurements and EQ based on GRAS could work better, on person B based on Neumann could work better etc, so in the end the only way to equalize tonally really good is by individual listening tests.

 

[flipflop]

Measured and EQed with a different ear they might sound more similar than before but there won't be a great match, especially at higher frequencies.

Without numbers, this can quickly become a discussion about semantics. What does "almost exactly the same" or "great match" actually mean in quantifiable terms?
Below is a table from the paper on EQ and distance perception I mentioned earlier in this thread.

As you can see, there's "No significant difference between Non-individual and Individualized equalization" when it comes to the subjects' performances.

 

[solderdude]

If you were to listen to the exact units measured and corrected by oratory, I think you would find them to sound extremely similar and very neutral as Olive's research would indicate. Are you suggesting this is wrong?

That is exactly what I am asking to forum members. Not what a specific measurement equipment tells us (after correction they are all 'flat' but ONLY on the rig where it is measured on, not on other rigs and not to people.

Also I am not judging OW research, not bashing on Oratory or others. I am asking members what their experiences are. Not what 'science' tells us.

If you take a headphone and apply the same EQ to it with 2 different pieces of software, it should sound the same unless there's something wrong with the at least one of the programs used.
Maybe I'm not understanding your question correctly.

Again I am NOT talking about EQ software sounding different with the same settings. They should not (but can) sound different when high Q filters are used that have pre-ringing or only post ringing. That's NOT what I am asking.

I am asking how different the end result EQ from Sonarworks (which is software) for instance is compared to EQ settings by Oratory or Jaakko or someone who tinkered with R'tings, or Griessinger, EARS or , heaven forbid, my own EQ.
What I am asking is how 'equal' a few headphones sound AFTER EQ is applied from different EQ suggesting (or delivering) settings or programs.

Yes, different measurement rigs will give different results. That's why it's a good idea to use the GRAS 45CA if the goal is to equalize headphones to the Harman curve.

I am not talking about target curves. It is about the spread in measurement accuracy and needed compensation. It is fine to bombard the GRAS 45CA as a standard. I am quite certain the 45CA coupler does not have the same properties as ears of a random person.

Sonarworks are notoriously secretive. The only thing I know is that they use a variation of the HTC, which seems to have too much energy in the bass and around 2 kHz.

It's not about the secrets nor their method. They arrived at their compensation based on comparing monitors to headphones and measurements.
That's all very fine. R'tings also made their own correction plot but now (for uniformity sake ?) chose to use the Harman curve but probably 'translated' to their coupler.
What I am asking is whether or not members with 3 or 4 different headphones and the 'settings' set to 'flat' perceive the compensated headphones to have the same tonal balance.
And what I asked is whether or not the spread in tonal balance is bigger than for instance Oratory.
Sure.. I could test this myself and be busy for a good few weeks to find out. I am sure though that many folks have tried this.
I don't care what Olive/Welti found.. just asking members for their experiences.
NOT about the theory behind it all or explanations... just experiences from members, first hand, not from papers or hear say.

Right, but they HATS is different from the one used by Harman, so the curve has all sorts of quirks to compensate for that fact.
To be exact, their correction curve goes all the way up to 20 kHz, but they don't weigh frequencies higher than 9 kHz for their scores, because the measured response is unreliable past that point.

Yes, again. I know all that. It also is not what I am asking.

Science is an ongoing process.

of course it is.

As demonstrated in the latest paper, both untrained and trained listeners preferred the Harman target over any of the virtualized headphones.

Again this is NOT about curves and preferences at all.
It is about the Delta in tonal balance between various headphones with applied 'compensation settings/programs' and if, in their experience, some members that experimented with different 'compensation settings' or Sonarworks found that one particular program or EQ settings had the smallest Delta between tonal balance.

 

[solderdude]

I've tried quite a few and so far oratory's settings using APO/Peace come closest to what I hear sitting in front of my calibrated nearfield monitoring setup playing the same content. I trust mixes I make using my 6XXs to translate to speakers very well in overall tonal balance.

This is the info I am looking for. Personal experiences with EQ settings or programs. The bonus is the correlation to monitors.
It is expected that both Sonarworks (which I assume went the same route) as well as O/W and Oratory have a similar approach and thus good results.

IMO the HD600/HD650/HD6XX are the headphones that need the least compensation and thus are already well tonally balanced from the start.
The less you need to compensate the better the sound.

 

[solderdude]

The problem is that is part of this HRTF is bypassed when listening to headphones, thats why EQing them doesn't work as well as it does for loudspeakers.

Exactly, this is true when comparing the sound of speakers with headphones but becomes as good as irrelevant when comparing headphones.
As good as because driver size, geometry, angle and driver-ear distance as well as acoustic feedback from the rear of the driver and the ear/pad cavity also have some influence.

The strange part (for me) is that when I listen to 'flat' nearfield monitors at a certain distance and angle sound very similar to headphones with a flat response.
Much less so with speakers in a room at listening distance and smaller distance between speakers.
Most mixing is done with monitors at relatively short distance in not so echoic circumstances with the speakers quite far apart. (there are exceptions)

From my few personal experiences and from what I have read from similar experiences in various forums there is not one measurement rig that works as well for everyone, so for person A measurements and EQ based on GRAS could work better, on person B based on Neumann could work better etc, so in the end the only way to equalize tonally really good is by individual listening tests.

This and it may be complicated by the applied compensation and how the manufacturer arrived at that compensation.
It is the compensation for all the resonances (which change with direction and relative driver size/distance) that leads to measured differences that are not compensated correctly. After all the compensation is just for one situation and obtained by averaging measurements.

I suspect that when one measures with a certain type of FR and phase distortions and then 'compensates' this with an 'average' compensation one measures peaks and dips that may not be there in reality. Those are the differences between couplers/compensation. One may be 'compensating' with EQ settings what should not have been compensated in the first place and while improving tonality may have introduced other errors.

I read that Oratory also listens (and perhaps fine tunes) EQ. I do this as well for obvious reasons.

 

[Kouioui]

650/HD6XX are the headphones that need the least compensation and thus are already well tonally balanced from the start. The less you need to compensate the better the sound.

So much this! That's why I use AKG 371s when I can't EQ or need isolation as they're Harman target tuned out of the box and play great from anything with a headphone jack. That said, I prefer them with oratory's settings too but the EQ'd 6XXs have other qualities that sets them apart from the rest.