Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Shouldn't all headphones be aggressively EQed to match HRTF? (Read 21048 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Shouldn't all headphones be aggressively EQed to match HRTF?

I've been very positively surprised by the effects of equalizing (using a software equalizer) headphones to get a flat perceived frequency response. There are some tutorials to do this, like http://www.head-fi.org/t/413900/how-to-equalize-your-headphones-a-tutorial . The effects of how perception of sound is different from the measurements is described by the HRTF model, so to some extent it could be applied by headphone manufacturers.

After equalizing Beyerdynamic DT880 the sound is much more realistic and mostly devoid of individual character. There's no "adaptation" period, frequently needed when changing headphones or starting a listening session - it sounds realistic from the beginning. It's hard to listen to un-equalized headphones now.

It generally comes to the 2kHz-8kHz range being too loud as much as 10-15dB - with rare exceptions like Audeze LCD-3. I know that the adaptation and "postprocessing" capabilities of a brain are very big, so these big peaks can be "removed" from consciousness. And the sound can come as more "detailed" and "energetic".

So, after my own experiments and reading about HRTF it looks like this issue is an elephant in the room - it has a huge effects on sound but is ignored by the industry. Am I missing something?

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #1 –
Using EQ to flatten out imperfections in headphone response has a significant (and in my experience positive) impact, although a little tricky to say everyone should do since headphone preferences vary so widely .

It's also relatively simple and very cheap so I doubt manufacturers want to talk about it.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #2 –
The problem is, these are not imperfections - there are massive differences between what 98% headphones produce and what HRTF tells is perceived. Like if the HRTF is ignored. Almost all headphones play 2kHz-10kHz too loud by a massive 10-20 dB.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #3 –
Quote from: artur on
http://www.head-fi.org/t/413900/how-to-equalize-your-headphones-a-tutorial
Why author thinks that he knows HRTF better than people fom headphone.com? Does he understand what he is doing? Does he know what reference for frequency response of headphones should be?
Quote from: artur on
Am I missing something?
Yes. Look here - http://rinchoi.blogspot.ru/2010/05/headphone-equalization.html
Of course for every person HRTF is individual and not the same as HRTF of dummy head. But to do meningful equalization you should know your own HRTF at first.

Quote from: artur on
After equalizing Beyerdynamic DT880 the sound is much more realistic
Much more realistic in comparison with what? What was reference for most realistic sound?

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #4 –
Sorry, this is going to be a short rant ...

First the bad:
a) Typical audiophiles fear signal processing. They sometimes even fear digital audio itself. Imho both stem from a complete ignorance about these topics, as well as incompetence ("I've put an EQ slider to +10 dB and now my sound is distorted, clearly EQ is evil" kind of nonsense). To these people the most important feature is a bypass button so I'm not surprised if manufacturers don't implement proper equalizers or sometimes even abandon such features.

b) EQ guides with sweeps where you correct for resonances do just that, they correct for individual resonances. They do not tell you anything about spectral balance, or what is "wrong" with a sound signature. Even worse, some of them do tell you some nonsense about how tones at different frequencies have to be played at the same loudness, which I will call the equal loudness trap. I know these people mean well and want to help, but this misinformation does more harm than good.

c) Newcomers use the wrong tools, then blame EQing while not being willing to invest some of their time to learn what EQs to use and how to use them. For example, taming a resonance with an easy-to-use 5 band EQ will usually produce a very dissatisfying result - again more harm than good.

d) People don't know about FF, DF .. equalizations, why equal loudness is simply the wrong target, HRTFs, or even what a "properly" equalized stereo system sounds like. Yes, the human hearing is quite flexible and can get used to a lot of different situations, but that also means that there are people used to systems that have abysmal sound. It gets worse when you look at the price tags of the components..

e) Confusion about how headphones operate ... "differently" at low vs. high frequencies. Break the seal with an in-ear headphone and you should see what I mean.
That there are individual differences in anatomy that lead to more or less different sound depending on the construction of the headphone, so what sounds perfectly equalized for me may not sound so good for you.
That there are manufacturing tolerances and differences between different headphones of the same model even if they're from the same batch, between the left and right headphone driver even which cannot be fixed if you spent thousands of dollars on audiophile tweaks but are easily fixed with equalization, which leads me to the good part..

edit: later
"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #5 –
Quote from: Rollin on
Quote from: artur on
Am I missing something?
Yes. Look here - http://rinchoi.blogspot.ru/2010/05/headphone-equalization.html

So headphones have consciously implemented  upward peaks in midrange/high freq. to compensate for ear gains?

Is my method for EQing wrong? I'm playing a Youtube video with a 20-20khZ sine wave: https://www.youtube.com/watch?v=qNf9nzvnd1k  and trying to achieve sound with the same loudness (of course it's largely subjective and requires multiple approaches).

What I'm confuesed about is that the article tells that compensation for ear gains should be upward peaks in HF, whereas my method (as well as almost all other from the linked head-fi thread) results in (sometimes very big) downward peaks in HF.[/quote]

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #6 –
Quote from: artur on
So headphones have consciously implemented  upward peaks in midrange/high freq. to compensate for ear gains?
Yes.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #7 –
Quote from: artur on
Is my method for EQing wrong?

To put it bluntly, yes. See my points b) and e).

And I'd like to add another point:
f) Equalization does have its limits. Digitally, electrically and mechanically/physically. A headphone that rolls off bass due to leakage (see again point e) will not like big bass boost. Output will not rise linearly. Distortion will rise dramatically.
"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #8 –
Quote from: xnor on
Quote from: artur on
Is my method for EQing wrong?

To put it bluntly, yes. See my points b) and e).

I only see arguments about why the equalization is not "the" solution for fixing all problems with sound. Sure. But I think it still works wonders in a lot of cases.

Isn't it generally impossible to freely adjust frequency response just by altering a headphone driver design + cups acoustics? I think you get "some" FR with can be roughly tuned (like applying felt on drivers), but the precision you get with a parametric EQ is impossible to achieve that way. Hence all headphones have big peaks in FR.

What's very surprising to me is that almost all people adjusting headphone FR end up lowering the 2-8kHz range by a big value like 10-20dB. This suggests that almost all headphones have a very big flaw - upper midrange / lower higher frequencies are too loud. I'm guessing this isn't "designed" but just impossible to fix without an EQ, which is not acceptable for an average audio user, so it's just left that way. Brain's accomodation works wonders.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #9 –
Quote from: artur on
I only see arguments about why the equalization is not "the" solution for fixing all problems with sound. Sure. But I think it still works wonders in a lot of cases.
Me too.
I haven't had time to write up the good part, but it's essentially this: frequency response is really important, no headphone can get it right, and equalization is the tool that only costs time but can result in tremendous improvements.

In short, EQ is a wonderful tool. Yes, you definitely should EQ your headphones. Everyone who isn't is missing something imho.


Quote from: artur on
Isn't it generally impossible to freely adjust frequency response just by altering a headphone driver design + cups acoustics?
Exactly. That's the first line of limitations that I've mentioned above.

Quote from: artur on
I think you get "some" FR with can be roughly tuned (like applying felt on drivers), but the precision you get with a parametric EQ is impossible to achieve that way. Hence all headphones have big peaks in FR.
The reason you get peaks in the FR is that at high frequencies headphones don't operate like "pressure chambers" anymore  (see point e). Instead, you will get reflections and resonances, such as the ear canal's.

Quote from: artur on
What's very surprising to me is that almost all people adjusting headphone FR end up lowering the 2-8kHz range by a big value like 10-20dB. This suggests that almost all headphones have a very big flaw - upper midrange / lower higher frequencies are too loud. I'm guessing this isn't "designed" but just impossible to fix without an EQ, which is not acceptable for an average audio user, so it's just left that way. Brain's accomodation works wonders.
I don't think so. 2-4k usually does not need to be adjusted that much in good headphones. Above that you will usually get multiple peaks. Other than that it's just an adjustment of low/mid/highs balance if your headphones have a smooth FR to being with.

You may end up with that result if you use a sweep or tones to equalize your headphone, yes, the equal loudness trap as I've called it above. But this is simply the wrong way to do equalization regardless if you use speakers or headphones actually.
"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #10 –
Quote from: xnor on
You may end up with that result if you use a sweep or tones to equalize your headphone, yes, the equal loudness trap as I've called it above. But this is simply the wrong way to do equalization regardless if you use speakers or headphones actually.

Why exactly is equal loudness "a trap"?

Quote from: xnor
EQ guides with sweeps where you correct for resonances do just that, they correct for individual resonances. They do not tell you anything about spectral balance, or what is "wrong" with a sound signature.

What is "spectral balance" and "sound signature"?

One problem I see is that you can get "better than real life" frequency response, because you can compensate for individual anatomical differences.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #11 –
Quote from: artur on
Why exactly is equal loudness "a trap"?
I'm not sure why people that discover the equal loudness contours it are so attracted to it, but I think their reasoning goes like this:
A headphone's frequency response has peaks and dips, therefore these frequencies will be played too high or low in volume, therefore correcting for equal loudness contours fixes all issues.

But that is simply wrong. A sound source that produces a flat frequency response will result in everything but a flat frequency response at your eardrum. For example, your ear canal can easily do a +10 dB boost between 2-3 kHz and this is perfectly natural and desired. Your pinna, head and even torse will have a significant effect too. Additionally, our hearing is also most sensitive in that range so if you compare sine tones it may very well sound 15 dB louder, but that's supposed to be that way.

If you attenuated that range by 20 dB then you could easily be 15 dB off. Equal loudness has a couple more problems, such as being SPL dependent or that it's based on artificial pure tones... Music has a way more broadband spectrum, noise components and also roughly a pink noise shaped spectrum (that is the spectrum drops by 3 dB each octave or doubling of the frequency).

Quote from: artur on
What is "spectral balance" and "sound signature"?
The balance between bass, mids and highs and the way this sounds. Boosted bass and attenuated highs will give you a dark sound signature. Rolled-off bass and boosted highs will give you a bright sound signature.
Boosted bass and highs will give you a loudness effect, also called a bathtub like response.


Quote from: artur on
One problem I see is that you can get "better than real life" frequency response, because you can compensate for individual anatomical differences.
I think you mean better than stock, unmodified. Yes, you can achieve that if you do it correctly. You can also completely mess up the frequency response and end up with a equal loudness curve equalized *boom-tish* (no mids) monster. :P
"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #12 –
Quote from: artur on
Why exactly is equal loudness "a trap"?

I think the largest part of what you are missing is this:

We all know and agree that a flat frequency response out of a transducer is desired.

But by (effectively) measuring frequency response at your eardrum (through your method) you're in effect canceling out all the ear responses which would still be present in the perception of a flat speaker's output.  You are not making the headphone flat, you're making the headphone-ear system flat, and that is not the appropriate goal.

And, let's face it, if your methodology results in a significant EQing down of 10-20dB @ 2-10K you're doing nothing but adding a smiley face EQ by another name - and a boost in bass and highs is well known to be pleasing, even though far from flat.
Creature of habit.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #13 –
Quote from: xnor on
Quote from: artur on
Why exactly is equal loudness "a trap"?
I'm not sure why people that discover the equal loudness contours it are so attracted to it, but I think their reasoning goes like this:
A headphone's frequency response has peaks and dips, therefore these frequencies will be played too high or low in volume, therefore correcting for equal loudness contours fixes all issues.

But that is simply wrong. A sound source that produces a flat frequency response will result in everything but a flat frequency response at your eardrum. For example, your ear canal can easily do a +10 dB boost between 2-3 kHz and this is perfectly natural and desired. Your pinna, head and even torse will have a significant effect too. Additionally, our hearing is also most sensitive in that range so if you compare sine tones it may very well sound 15 dB louder, but that's supposed to be that way.

You are talking about measured frequency response flatness, and the mentioned equalization methods give perceived FR flatness - they, by definition, give "flat frequency response at your eardrum".

Quote
Additionally, our hearing is also most sensitive in that range so if you compare sine tones it may very well sound 15 dB louder, but that's supposed to be that way.

If a sine wave is recorded at a constant loudness then a perfect audio system should produce a perceived constant loudness?

Quote
Quote from: artur on
One problem I see is that you can get "better than real life" frequency response, because you can compensate for individual anatomical differences.
I think you mean better than stock, unmodified.

No, I meant real life. The sound of instruments or songs was created and tuned for an average person, so if your hearing differs, the frequency range you actually hear will be different from other people. But by equalizing using sine waves, you can hear how other people do it. It's like disability equipment.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #14 –
Quote from: Soap on
We all know and agree that a flat frequency response out of a transducer is desired.

As @xnor wrote, no, you don't want measured flatness.

Quote
And, let's face it, if your methodology results in a significant EQing down of 10-20dB @ 2-10K you're doing nothing but adding a smiley face EQ by another name - and a boost in bass and highs is well known to be pleasing, even though far from flat.


2-10K is not midrange, it's treble, so it's mostly cutting out high frequencies. The initially perceived sound difference is that it's "analog" or even "muddy" - very different from a "loudness" effect. Also, if you are doing it precisely, there will be a lot of deep peaks so it's not a smooth "smile". For example, for Grado SR60 a difference I got between a 2.5kHz and 3.5kHz band is 18dB.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #15 –
Quote from: Soap on
We all know and agree that a flat frequency response out of a transducer is desired.

But by (effectively) measuring frequency response at your eardrum (through your method) you're in effect canceling out all the ear responses which would still be present in the perception of a flat speaker's output.  You are not making the headphone flat, you're making the headphone-ear system flat, and that is not the appropriate goal.
It's a typo? What you try to get flat is the "source of the sound" since you want to reproduce a recording in the most similar way it was originally recorded; i.e. what you would hear if you would have been there. The source of the sound changes according to the situation.

Using speakers you need to flatten the spekears response within your listening room.

Using headphones you need to flatten headphone within your eardrum.

Both situations need to take into consideration some type of room correction, let it be a real room or you ear. Headphones already are supposed to account for that correction at some level, another thing is they introduce other peaks and undesired coloration...

Speakers are easier to correct, headphones require much more work. Both are possible to correct at some extent and yes, obviously they don't put more interest on this because that gives no money.

Flatten the freq. response in this context is not equal to make all freqs. coming from the headphone or the speakers sound equally loud. The reference is the original recording and you compare that to what you hear; what you flatten or minimize are the differences between both not an absolute value. White noise must sound like white noise in your head or 4m in front of your speakers, you got HRTF to measure the first and a microphone for the second.

The white noise example is pretty simple since even using speakers you probably need to apply some room corrections so the sound coming from the speakers is not a perfect flat line when you perform a spectral analysis but that's what you aim/get at the listening point. Flatness is a term relative to the listening place. Flatness out of a transducer is probably not going to be faithful to the reference at the listening point.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #16 –
Quote from: artur on
You are talking about measured frequency response flatness, and the mentioned equalization methods give perceived FR flatness - they, by definition, give "flat frequency response at your eardrum".
No.
My example used a flat FR only as a reference, to explain that what arrives at the eardrum will NOT be flat and what you will hear is yet again different - even less flat - because your hearing is simply a whole lot less sensitive at 20 Hz than it is at 2 kHz. This applies to any arbitrary frequency response, but I now see that mentioning a "flat FR" probably just added to the confusion.
Again, equal loudness equalization is wrong. It will not give you anything flat, except for a perceived flatness of single, pure tones. This is wrong.

Imagine an artificial instrument that produces a 100 Hz and 3 kHz tone at 60 dB SPL. To make this simple, we assume we are in a free field (there is no room). Let's say you will perceive those tones as roughly 40 phon and 65 phon. That is our natural starting point.
Now we record this instrument, play it through flat speakers, again in free field conditions. Same perceived result - everything is fine.

Now you equalize for equal loudness, so +20 dB at 100 Hz, -5 dB at 3 kHz. Now effectively the recorded instrument produces the tones at 80 and 55 dB SPL instead of 60 and 60. You've completely messed up the FR.

In reality it's not that simple.

Quote from: artur on
If a sine wave is recorded at a constant loudness then a perfect audio system should produce a perceived constant loudness?
I don't see how this makes much sense.
A sine wave is usually defined to have a constant amplitude. We record sound pressure, not loudness. Loudness is what we perceive.
"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #17 –
Quote from: artur on
No, I meant real life. The sound of instruments or songs was created and tuned for an average person, so if your hearing differs, the frequency range you actually hear will be different from other people. But by equalizing using sine waves, you can hear how other people do it. It's like disability equipment.
I think I understand what you're trying to say, that individual calibration is better, but music is neither produced that way nor does your point make sense in the light of loudness equalization of your headphones with pure tones (which is wrong).
Instead, you would have to equalize your headphones and each track individually.

Quote from: artur on
As @xnor wrote, no, you don't want measured flatness.
"Measured flatness" is meaningless for headphones.

"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #18 –
Ok, so it seems the correct way for equalization would be to have pure tones compensated for hearing differences at a specific sound pressure, according to measurements:



And then aiming for the same loudness?

Ok so it looks like with pure tones I got the midrange too quiet, but still the method is valuable as headphones have huge peaks, exceeding the expected values from the graph.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #19 –
One more thing...
Equal loudness equalization, what the OP did, is not "measuring frequency response at your eardrum" and making it flat. That is simply equalization at the eardrum reference point (DRP).
A 20 Hz and 200 Hz tone produced at some SPL can arrive at almost equal SPLs at the ear drum, but the 20 Hz tone will be perceived as much lower in loudness.

Another measurement point is the ear entrance. But don't worry, all of those are wrong when equalizing for a flat FR.
"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #20 –
Quote from: artur on
Ok, so it seems the correct way for equalization would be to have pure tones compensated for hearing differences at a specific sound pressure, according to measurements:



And then aiming for the same loudness?

This is what I've usually done.  It works fairly well (to my ears), although it's never been clear to me if the use of pure tones is actually correct. 

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #21 –
Quote from: artur on
Ok, so it seems the correct way for equalization would be to have pure tones compensated for hearing differences at a specific sound pressure, according to measurements:



And then aiming for the same loudness?

Ok so it looks like with pure tones I got the midrange too quiet, but still the method is valuable as headphones have huge peaks, exceeding the expected values from the graph.

First of all, you are using an outdated measurements from Fletcher and Munson. Use ISO 226:2003 instead:




Secondly, if I understood you right you now want to do the reverse: an equal SPL equalization based on equal loudness contours?
Keep in mind that:
Above loudness contours are averages. Your hearing will be off by a few dB. Add this to the impreciseness of matching tones by loudness. (Maybe cross checking with multiple equal loudness curves helps?)
Also, the result will not sound great or "right" just like a speaker equalized flat at the listening position in a room. You will probably want to add a little bass boost and some downward slope of the FR.
"I hear it when I see it."

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #22 –
Quote from: Isabelxxx on
It's a typo? What you try to get flat is the "source of the sound" since you want to reproduce a recording in the most similar way it was originally recorded; i.e. what you would hear if you would have been there. The source of the sound changes according to the situation.
Correct.

Quote
Using speakers you need to flatten the spekears response within your listening room.

Correct.  Which means a non-flat response @ your eardrum due to the effects of your outer ear, ear canal, etc.

Quote
Using headphones you need to flatten headphone within your eardrum.

Incorrect, because now you've canceled out the previously mentioned modifiers.

The issue here is one of the methodology of trying to make heard tones all the same volume.  That's using the eardrum, whereas if you did it properly in a room you'd use a microphone which does not compensate for the unique-to-you modifications to FR of your personal ear geometry.
Creature of habit.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #23 –
Quote from: Soap on
Incorrect, because now you've canceled out the previously mentioned modifiers.

The issue here is one of the methodology of trying to make heard tones all the same volume.  That's using the eardrum, whereas if you did it properly in a room you'd use a microphone which does not compensate for the unique-to-you modifications to FR of your personal ear geometry.
I get what you mean; correcting speakers you are listening to the original source without accounting for anyone's ear correction, the way It should be.

While ear's correction introduces a further modification not present in the original source. But that's just a confusion of what needs to be corrected and what not.

Maybe I did not use the right terms, what I consider as flattening the headphone FR within your eardrum is doing the same than the speaker analogy! Maybe we need to be more careful about the terminology here.
It's clear a correction needs to be made but you don't need to go so far to the point of correcting the natural modification your ear makes to any sound, let it be a speaker or a headphone... I'm only speaking about a correction accounting for the source being within your ear, not the other corrections. Clear now? The same the speaker placement (more or less near your) affects the correction needed but for ex. you don't go further adding your own hearing losses to get a flat response compared to the original source. Neither you try to make all tones sound to you equally loud!

There speakers correction seems to be clear to anyone and you only need to translate that idea to headphones having in mind new corrections may be needed and others factors may be considered to not take away natural modifications you don't touch with speakers correction.

Any sound, the one coming from the speaker too, is modified in my ear but that does not need to be corrected. You can do the same with a headphone having in mind you only need to account for the speaker being in your ear but not going further to make all tones been heard at the same volume, which is not the natural way of hearing things.

Just to be clear, independently of the terminology, the idea is to hear exactly what you/microphone would hear in the recording room. If you correct a headphone FR you should hear the same than using a perfect speaker with flat FR at the listening point. If both listening matches then that's the right correction. Now the difficult point is that... you can not easily compare a headphone FR with a speaker FR since the first needs to be measure by the listener... and you can not easily ABX both sources.

The equal loudness curves use is not right having in mind that idea and it's easily noticeable if you ABX headphones corrected (that way) vs speakers with flat FR. While not perfect for fine tuning it should be easy for anyone to note the 1st correction is not sounding the same than the second. You don't naturally hear all tones equally loud and a speaker corrected properly shows that, the same you would note being in the recording room.

In fact I think we agree in all points Soap.

Re: Shouldn't all headphones be aggressively EQed to match HRTF?

Reply #24 –
Not "in" your ear. "By" your ear would get you closer as the pinna plays a big role on FR.

Ironic, reading the title which is also something one needs to consider which isn't exactly about this tangent that has dominated the conversation.

But, sure, let's digress a little bit...
Here's something to consider about equal loudness contours which has far more to do with trying to replicate what was heard during mastering:  what was the level of playback at that time?  People spend so much time worrying about the dips and peaks but seem not to notice how they flatten with increasing level. Unlike the other (dips and peaks), this has relevance.

So, HRTF anyone?