Is EQ'ing headphones worth it?

[solderdude]

Of course EQing has a significant effect on preference ratings. That is the whole point of EQ.

That does not mean that you can take a HD201 and make it sound as good as an HD800.
You can give it a similar tonal balance but it won't sound the same.
When you can, please post the EQ and you can save a lot of people shitloads of money.

I have EQ'ed a few headphones to closely the same target and while tonally they sound quite similar there is a distinct difference in treble quality to me. That said, I am sure if i would let some family members listen to the headphones they would say they sound the same and all sound good.

EQ can be (actually is) beneficial when done properly but there are limits to what one can do. I also am convinced that it takes a lot of experience in listening to be able to tell differences others (untrained) cannot even hear. Once you hear some of these things they are hard to unhear.

 

[Blujackaal]

Oh right. We have eq!

Add another 5db you got the ER4B treble(ignoring other factors) which i like more than just 5db(ER4S) alone imo.

 

[Erik]

Of course EQing has a significant effect on preference ratings. That is the whole point of EQ.

That does not mean that you can take a HD201 and make it sound as good as an HD800.
You can give it a similar tonal balance but it won't sound the same.
When you can, please post the EQ and you can save a lot of people shitloads of money.

I have EQ'ed a few headphones to closely the same target and while tonally they sound quite similar there is a distinct difference in treble quality to me. That said, I am sure if i would let some family members listen to the headphones they would say they sound the same and all sound good.

EQ can be (actually is) beneficial when done properly but there are limits to what one can do. I also am convinced that it takes a lot of experience in listening to be able to tell differences others (untrained) cannot even hear. Once you hear some of these things they are hard to unhear.

What makes you think that you can't take a HD201 and make it sound as good as an HD800? Equalized Beats or Bose headphones did sound as good as the highest end Stax. So why the HD201 can't sound as good as the HD800? I suspect that you are making an anecdotal assumption here.

No one has said that two headphones will sound absolutely the same after equalization to the same target. According to the study I linked, there will remain some negligible difference, but it won't be enough for a listener to confidently say which headphone sounds better.

Have you used in-ear microphones to match the headphones to the same target? Otherwise they were not matched to the same target when you compared them. Was it a controlled, blind comparison? Otherwise you were biased and that affected your experience.

 

[MRC01]

What makes you think that you can't take a HD201 and make it sound as good as an HD800? Equalized Beats or Bose headphones did sound as good as the highest end Stax. So why the HD201 can't sound as good as the HD800? I suspect that you are making an anecdotal assumption here.
...

In theory, distortion and CSD differences can be audible even when FR is identical. In practice, experience supports this. For example if I EQ my HD580 to have perfectly flat bass response down to 20 Hz, like my LCD-2F does naturally, their bass doesn't sound the same. The LCD-2F is crisper, cleaner, tighter. The HD580 in comparison is bloated and wooly after being EQed to the same level. The difference is not subtle. Nor is it limited to the bass. Distortion as low as 1% can be audible in the midrange and treble, and CSD / ringing can also be audible. And EQ (especially boosts) exacerbate distortion and CSD.

In short, I agree that frequency response accounts for most of the differences we hear. But it doesn't account for all of what we hear.

 

[Erik]

In theory, distortion and CSD differences can be audible even when FR is identical. In practice, experience supports this. For example if I EQ my HD580 to have perfectly flat bass response down to 20 Hz, like my LCD-2F does naturally, their bass doesn't sound the same. The LCD-2F is crisper, cleaner, tighter. The HD580 in comparison is bloated and wooly after being EQed to the same level. The difference is not subtle. Nor is it limited to the bass. Distortion as low as 1% can be audible in the midrange and treble, and CSD / ringing can also be audible. And EQ (especially boosts) exacerbate distortion and CSD.

In short, I agree that frequency response accounts for most of the differences we hear. But it doesn't account for all of what we hear.

Have you missed my first post in this thread? I already posted a link to the study about non-linear distortion audibility and its effect on listeners' preference in headphones. It says that in practice non-linear distortion is not a significant factor in subjective sound quality evaluations as long as there is nothing abnormal with it. That contradicts your anecdotal experience.

 

[Dreyfus]

My observations are quite the same as solderdude's.

I did a few comparisons in the past, matching phones like the Superlux 681, Beyer DT 990, Sennheiser HD 650 and HD 700, Audio Technica R70X, Audioquest Nighthawk and Massdrop HE4XX. Some with manual loudness matching by ear (1-2 hours per headphone), some with in ear measurements and some with data from Oratory, IF or rtings.
In most cases they sounded really similar indeed, but not identical. My guess is that it is the inconsistency in the higher frequencies (hard to nail moving peaks and troughs) and the actual physical build that alter the sound. Some open back designs may increase spatial depth and reduce in-head localization due to crossfeed.
There is always room for improvement when talking about the measurement / matching techniques, of course. But I doubt that such effects can sufficiently be reproduced by simply tweaking the frequency response.

Sometimes I feel like every composition - different inner and outer shapes, different pads and materials, different coupling angles and distances etc. - has its own distinct sound that cannot be reproduced by FR matching. That may be complete audiofool bs in the perceptible practice. But the fact, that all those hours of work did not deliver pleasing results so far, grew the impression in me that there must be something to it.

Maybe Harman wanna send me an invitation?
I would be delighted to prove or falsify my impressions.

 

[MRC01]

Have you missed my first post in this thread? I already posted a link to the study about non-linear distortion audibility and its effect on listeners' preference in headphones. It says that in practice non-linear distortion is not a significant factor in subjective sound quality evaluations as long as there is nothing abnormal with it. That contradicts your anecdotal experience.

I read your link and my experience is not anecdotal. The notes I shared are from my own ABX testing adding different levels of distortion to music to see at what point I can detect it.

 

[pwjazz]

Have you used in-ear microphones to match the headphones to the same target?

As a side note, I've played with this and it's hard to get a good result with this too. Either you have a little mic that sits in the concha, which will affect concha gain, or you occlude the ear canal with the mic which changes your ear's acoustic impedance

 

[bobbooo]

I read your link and my experience is not anecdotal. The notes I shared are from my own ABX testing adding different levels of distortion to music to see at what point I can detect it.

Could you describe how exactly you did these ABX distortion tests? I've been wanting to do a similar test myself for a while. What did you find was the lowest percentage distortion you could detect? And I presume you made sure to do enough trials to achieve at most 1% probability of success by chance (e.g. 13/16 correct guesses)?

 

[Asylum Seeker]

View attachment 50628
And to people who don't exactly have much idea about how two targets compare. For the most part they are more similar than different.

Biggest difference ia the <200Hz región. How does the Etymotyc target jibe with Fletcher Munson?

 

[JohnYang1997]

Biggest difference ia the <200Hz región. How does the Etymotyc target jibe with Fletcher Munson?

The bass is based off preference which I cannot argue with. Also the bass does not change tonality.
On the Fletcher Munson curve. It should not be used in any way for target curve. It can be used for eq. But the part that 3-4khz is recessed should not be accounted. Use this graph instead: (the blue curves)

 

[solderdude]

What makes you think that you can't take a HD201 and make it sound as good as an HD800? Equalized Beats or Bose headphones did sound as good as the highest end Stax. So why the HD201 can't sound as good as the HD800? I suspect that you are making an anecdotal assumption here.

No one has said that two headphones will sound absolutely the same after equalization to the same target. According to the study I linked, there will remain some negligible difference, but it won't be enough for a listener to confidently say which headphone sounds better.

Have you used in-ear microphones to match the headphones to the same target? Otherwise they were not matched to the same target when you compared them. Was it a controlled, blind comparison? Otherwise you were biased and that affected your experience.

Negligible to who ?
Did you participate in the study ?
Have you managed top make certain headphones sound equally good as an Utopia, HD800(S), K812, Susvara or other high end headphone yourself ?

I also have experimented with IEmics and the results were not encouraging at all.

No controlled, no blind tests (of course) even remotely THINKING that such is possible is laughable. Certainly when one has heard and felt a lot of headphones. An LCD2 does not feel like an HD800, HD650 or Focal. LCD2 vs LCD3 or so .. sure that could be blind.

And yes, I am biased and experiences sure are biased and was affected by it.

And yes, one could just use an HD201 and EQ it to a target or a target that has the tonal balance of another headphone with the flip of a switch and then have someone make notes which sound they prefer (tonally). That is NOT the issue but is what was tested a preference for a tonal balance so they could find out what people want so they could increase sales.

They never set out to see if an experienced listener can be happy with a $ 15.- headphone (many are).
For a long time I could have lived with an EQ'ed HD681 (in fact I did). Even though I liked it a lot it definitely did never reach the level of detail retrieval, pin point accuracy and overall sound quality as some of my better headphones (with EQ).

 

[thewas]

Sometimes I feel like every composition - different inner and outer shapes, different pads and materials, different coupling angles and distances etc. - has its own distinct sound that cannot be reproduced by FR matching.

Exactly, headphones interact differently with every different ear and measurement rig due to above reasons, so thinking it is possible to equalise them by just one measurement at one point is similar like thinking that 2 loudspeakers will sounds the same when equalised from one angle measurement which only works in reflexion free conditions and listened to the same angle. Also all kind of distortions (often KD is over and IMD is underestimated) should be of course lower than the perception limits which especially in bass is barely the case, its hard enough to EQ a HD6x0 flat in bass, not talk something cheaper.

 

[Erik]

Negligible to who ?
Did you participate in the study ?
Have you managed top make certain headphones sound equally good as an Utopia, HD800(S), K812, Susvara or other high end headphone yourself ?

I also have experimented with IEmics and the results were not encouraging at all.

No controlled, no blind tests (of course) even remotely THINKING that such is possible is laughable. Certainly when one has heard and felt a lot of headphones. An LCD2 does not feel like an HD800, HD650 or Focal. LCD2 vs LCD3 or so .. sure that could be blind.

And yes, I am biased and experiences sure are biased and was affected by it.

And yes, one could just use an HD201 and EQ it to a target or a target that has the tonal balance of another headphone with the flip of a switch and then have someone make notes which sound they prefer (tonally). That is NOT the issue but is what was tested a preference for a tonal balance so they could find out what people want so they could increase sales.

They never set out to see if an experienced listener can be happy with a $ 15.- headphone (many are).
For a long time I could have lived with an EQ'ed HD681 (in fact I did). Even though I liked it a lot it definitely did never reach the level of detail retrieval, pin point accuracy and overall sound quality as some of my better headphones (with EQ).

Neglible to the trained listeners who participated in the study. It does not matter if I participated in the study or not. And it does not matter if I managed to do something or not. We are not discussing my person here.

Good thing that you have accepted the fact that you are making anecdotal assumptions based on a biased experience. Next step will be to refrain from using such assumptions as a counterargument to evidence gained in controlled experiments.

The study I linked was not about a preference for a tonal balance. You are confusing it with other Harman publications. I will post its brief summary again under the spoiler and highlight the most interesting part.

Spoiler: A brief summary of the study

In the study I linked above, five headphones were equalized to the same target. The headphones were AKG K701, Beats by Dre, Bose QC15, Sony MDR-V600 and Stax SR-009. Conclusions of the study:
• Headphone had a significant effect on preference ratings; this effect was largely isolated to one (Headphone D), which was less preferred to the other four headphones. There were no significant preferences among the other four headphones.
• Listener comments indicated Headphone D had audible distortion. For the other headphones, listeners’ comments were often inconsistent across repeated observations, and expressed how difficult it was to hear audible distortion in the headphones.
As you can see, only one pair of headphones out of five sounded noticeably worse than the others. That means that either Beats or Bose headphones sounded as good as the highest end Stax when they were equalized to the same target.

 

[solderdude]

You forgot to highlight other interesting parts

Headphone had a significant effect on preference ratings; this effect was largely isolated to one (Headphone D), which was less preferred to the other four headphones. There were no significant preferences among the other four headphones.

I have no problems discussing anecdotal assumptions based on a biased experience as long as it is clear that it is just that.
Are you concluding that the participants in the test did things any different ?

For the other headphones, listeners’ comments were often inconsistent across repeated observations, and expressed how difficult it was to hear audible distortion in the headphones.

to get back on topic: Is EQ'ing headphones worth it?
My answer remains the same.. Yes in most cases but realise this may not be true for everyone due to differences in personal taste.
It is useful because in the study they too EQ'ed so the answer is yes acc. to the study as well.

Also products were made in accordance to these studies which shows the usefulness of measurements and research for preferences and a standard.
But these products too have shortcomings and do not follow the target exactly and need some EQ as well which indicates how difficult it is to make a good performing/sounding headphone. And those models also are not universally liked by all people.

So yes, EQ away if you want or need it and don't if you think it sounds fine.
Worth it to me ... hell yes... and based on my flawed (non compliant) measurements and subjective personal opinions as well.

 

[DivineCurrent]

After years of experimenting with EQ, and owning various popular headphones, I agree with Solderdude's experience.

No matter how much you EQ a Sennheiser HD 650 to have a flat measured bass response all the way down to 20 Hz, it will never sound as tight and controlled (and yes, in my opinion not as "good") as something like an LCD-2. Why? Well, take a look at the driver design. I don't claim to know much about physics, but planar magnetics have far more surface area to work with than the 40mm Sennheiser dynamic driver, and therefore can physically push more air with lower distortion. No matter what, you can't turn a 20+ year old dynamic driver design and change it into a 106mm planar magnetic driver with powerful magnets. EQing the sub bass to high levels on the 650 would be like taking a bookshelf speaker and trying to EQ the sub bass below 50 Hz and expect subwoofer performance. Not going to happen, it wasn't designed for that. On top of that, the open back Audeze LCD series in particular have better sealing to the head and trap the low frequencies so you can hear them better. Different driver designs can sound vastly different even with EQ, even though that may not be very apparent on FR graphs.

Of course, preferences differ and there are many low cost headphones that subjectively sound very good to me. I happen to prefer the $10 Sony MH755 earphones over the $125 Etymotic ER2SE, even if they are a technically inferior design.

 

[MRC01]

Could you describe how exactly you did these ABX distortion tests? I've been wanting to do a similar test myself for a while. What did you find was the lowest percentage distortion you could detect? And I presume you made sure to do enough trials to achieve at most 1% probability of success by chance (e.g. 13/16 correct guesses)?

Several have been posted here - sites that host a high quality recording, play it back in trials starting with high distortion, incrementally reducing with each trial. Search and you will find them, and try it yourself if you like. However, in testing trained listeners, audible distortion threshold is generally found to be around 3% in the bass, 1% in the mids and a bit lower in the treble. These distortion thresholds are reflected in the ITU standard for audio equipment used for blind testing (BS.1116, I believe).

The ideal confidence level of an ABX test depends on the goal of the test. An ABX test cannot have both high precision and high recall. High confidence percentiles are high precision but low recall; they reduce false positives but they increase false negatives. Low confidence percentiles (still above 50% of course) are low precision but high recall; they reduce false negatives but they increase false positives. This has been discussed at length on this forum, search and you'll find it.

 

[Robbo99999]

Thanks for everyone's responses so far, turns out I had a little time today so I've done my first 'proper' foray into EQ'ing my AKG K702's by downloading Equaliser APO and applying what I believe to be an "oratory1990 curve" from here: https://github.com/jaakkopasanen/AutoEq/tree/master/results
and this is a screen shot of the applied curve in my Equaliser APO:

I've done some back and forth testing of stock AKG K702 vs the curve above, and I guess the oratory1990 curve does sound more 'balanced' overall, but I feel that the equalised curve reduces speed & detail of female voices (don't know if speed is a term) - but I'm not getting all the nuances and quick changes within a female vocal. I specifically used Florence and the The Machine on their St Jude track listening to a section of it from about 3 mins into the track, and I tried to volume match by ear to remove as much as possible volume bias. So I'm fairly liking the general sound of the oratory1990 curve, but it's missing some detail in female voices....why is that....do you think the oratory1990 is boosting the base too much and therefore removing female vocal detail? I believe this is Harman target, so maybe I should try the above curve but with the bass 'removed' so it would be similar to Etymotic Target - that's my first hunch (and also influenced by a post on page 1 by @JohnYang1997 ).

This is just a first little proper stab at some EQ'ing, no more time today on it, but next week or Sunday I have some more time for a second attempt based on anything I learn & hear from you guys, thanks for all your inclusions in this thread.

EDIT: and John I haven't forgotten about this post of yours (https://www.audiosciencereview.com/...-eqing-headphones-worth-it.11523/#post-330487 ), just I'm trying a quick fix approach first with the limited time I have today.

EDIT#2: and that curve I applied in the pic above, it also seems to have removed detail and timbre of instruments from further down the frequency band (so not only removing detail from female vocals) - for example the intro (first minute) of Metallica's Enter Sandman track the detail and timbre of that is decreased and lost, and that's a low frequency part with the "power chord guitars".

 

[Kouioui]

I happen to prefer the $10 Sony MH755 earphones over the $125 Etymotic ER2SE, even if they are a technically inferior design.

The MH755 comes quite close to the Harman target without EQ. The ER2SE needs lots of help to get there. The 650/6XX and AKG K371 are closer to the Sony without EQ. The sub bass on the AKG is surprisingly strong and clean. The 6 series Senns will never match it but the mid-bass bloat can be removed.

 

[bobbooo]

Thanks for everyone's responses so far, turns out I had a little time today so I've done my first 'proper' foray into EQ'ing my AKG K702's by downloading Equaliser APO and applying what I believe to be an "oratory1990 curve" from here: https://github.com/jaakkopasanen/AutoEq/tree/master/results
and this is a screen shot of the applied curve in my Equaliser APO:
View attachment 51140

I've done some back and forth testing of stock AKG K702 vs the curve above, and I guess the oratory1990 curve does sound more 'balanced' overall, but I feel that the equalised curve reduces speed & detail of female voices (don't know if speed is a term) - but I'm not getting all the nuances and quick changes within a female vocal. I specifically used Florence and the The Machine on their St Jude track listening to a section of it from about 3 mins into the track, and I tried to volume match by ear to remove as much as possible volume bias. So I'm fairly liking the general sound of the oratory1990 curve, but it's missing some detail in female voices....why is that....do you think the oratory1990 is boosting the base too much and therefore removing female vocal detail? I believe this is Harman target, so maybe I should try the above curve but with the bass 'removed' so it would be similar to Etymotic Target - that's my first hunch (and also influenced by a post on page 1 by @JohnYang1997 ).

This is just a first little proper stab at some EQ'ing, no more time today on it, but next week or Sunday I have some more time for a second attempt based on anything I learn & hear from you guys, thanks for all your inclusions in this thread.

EDIT: and John I haven't forgotten about this post of yours (https://www.audiosciencereview.com/...-eqing-headphones-worth-it.11523/#post-330487 ), just I'm trying a quick fix approach first with the limited time I have today.

Oratory1990's EQ settings are for the AKG K702 65th Anniversary Edition. Do you have this edition? If not, and you have the normal edition, then you're using the wrong EQ settings, as the two editions have different tuning. This would explain why it doesn't sound right to you. Unfortunately for you, it doesn't look like Oratory1990 has measured the regular K702.

You could try these PEQ settings based on Tyll's measurements of innerfidelity.com. There are two caveats with these settings though. Firstly, the measurements by Tyll, although great, are not professional, and use an older HATS measurement system. Oratory1990 however is a professional acoustic engineer with an academic background in physics who specifically measures headphones as part of the research his company does based in Austria, using state of the art equipment. I don't know of any other headphone measurement database available online with this level of expertise behind them. The second caveat with the Innerfidelity EQ settings on AutoEQ is that they use a non-standard target curve by a user on this forum, which unlike the Harman target, has no scientific basis.

A final note: even the default AutoEQ settings that say they use the Harman target are not quite correct - they are 2dB and 4dB lower in the bass then they should be for the over-ear and in-ear settings respectively. Also, the AutoEQ readme says "Above 6 to 8kHz data is filtered more heavily to avoid measurement artifacts and no positive gain (boost) is applied. In the upper treble measurements are less reliable and boosting them too much will cause serious problems while having some narrow dips is not a problem at all." This is sensible, but you may achieve better results by relaxing this constraint. This, and increasing the bass to correctly match the Harman target, can both be done by downloading the AutoEQ program itself and using it to create your own custom EQ profiles by changing the parameter values it uses. Details can be found in the readme.