OP
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Is EQ'ing headphones worth it?
- Thread starter Robbo99999
- Start date
- Joined
- Oct 25, 2019
- Messages
- 11,094
- Likes
- 14,751
When you know, can you tell the rest of us. Preferably without graphs.
I'll try my best! I have read a few of the previous post made and I do agree that loudness matching makes sense as the way in which we perceive sound can change based on how loud you have a pair of speakers or headphones playing a certain song or noise. The tricky part I get is the matching part cause memory and patience can be a pain to deal with when doing the method Griesinger performed in his video. I have even made an Audacity file full of alternating pink noise bands like Griesinger has and it helps a bit but I'm still playing around with it. I'm working on EQ'ing my pair of headphones (Status CB-1) to an old but fairly nice sounding speakers (Magnavox LSB 300/15) that came with a old turntable my granny has.
OP
- Thread Starter
- #245
You asked me to explain this quote from me "Of course equal loudness equalization will result in unnatural colouration because our ears/brains do not perceive equal loudness across the frequency range.".
Even if you forget the Equal Loudness Graphs I've been posting I'm sure you're aware that your hearing sensitivity begins to tail off towards the bass end of the spectrum and towards the high end of the spectrum no matter if you use headphones or speakers....if you were to test yourself with test tones on a calibrated speaker or headphone then that is what you would experience, and you'd likely experience that on uncalibrated gear unless it had some crazy v-shaped EQ profile where the bass and high end were severely artificially boosted. So, yes, with that basic premise in mind if you were to equal loudness calibrate your speakers or headphones across the entire audible frequency range then of course you would end up unnaturally boosting the lows & highs....that's easy to visualise right?
The Equal Loudness Graphs show a bit more granularity/detail to that, and yes that is for speakers (I'm assuming!), but those graphs show the same phenomenon we're talking about in the previous paragraph, namely that hearing sensitivity changes across the frequency range. And as I've said before this implies that the only frequency range that is valid for Equal Loudness EQ is the portion of that graph that is flat, is that the part you're struggling to visualise?
Now we don't have those Equal Loudness Graphs for headphones (that I'm aware of), and I think base choices like whether a headphone should be EQ'd for HRTF or not come into play here too before we can decide if we could use a flat portion of any theoretical (not been created yet?) Equal Loudness Graph for headphones, because I think you'd be EQ'ing your headphones with no HRTF in mind if you tried Equal Loudness EQ on headphones, so that's one limitation I think. So I don't think I'm gonna talk anymore about Equal Loudness Graphs for headphones because it's not a known area and other factors come into play. But in general terms I think it's possible to postulate as to whether Equal Loudness EQ'ing is valid for headphones, and I already postulated on that in the first paragraph of this post, and I don't think Equal Loudness EQ'ing is valid given that reason alone.
When it comes down to it I think there are better ways of EQ'ing speakers and headphones. Speakers - use a UMIK / REW and a target curve of your choice. Headphones - use a target based on research and an EQ based on measurements from a proper rig.....headphone measurement is difficult and costs lots of money for the equipment if it's a good one....Oratory1990 has either one of the best or the best rig...so use his EQ settings which are also to the Harman Curve. Of course if you don't like the results you could use his raw data from the headphone measurements and then use REW to EQ that raw measurement to a modified target curve of your choice....that's probably the next best option in my opinion.
Yeah, ok ¯\_(ツ)_/¯
Its not unnatural if I perceive it as being tonally balanced with both noise and music. In the end, that's also what Harman did. Their curves could only be validated by actually letting subjects listen to that curves and adjusting the faders accordingly. Measurements and target curves without correlation to (subjective) human hearing and practical proving are of no real life use.
Yes, because it has little to do with the topic of subjective headphone equalization.
I suggest we totally skip on anything that includes the words "equal", "loudness" or "contour" and just talk about equalizing a headphone with a given set of bands to your liking until you subjectively perceive it as flat through the whole frequency band. That is what I was talking about all the time. I don't try to force any direct correlation to speakers, nor do I apply research that has been founded on listening to speakers in a (wet or dry) room. Listening to headphones which places the transducers right infront of your pinna is a totally differnt thing. There is no common standard for equalizing that particular reference point since it is not an established production standard. Hence, there is no right or wrong, only taste and subjective evaluation. And that is where the equalizer comes in which addressed a) irritating resonances which do not match your personal hearing and b) restores the tonal balance for the bass, mids and highs.
OP
- Thread Starter
- #247
Yes, we should leave it, I'm convinced you're skipping some elements of understanding on the points I'm pointing out so we're not getting anywhere, but we can beg to differ, I think some of the points are very obvious. And anyway, when it comes down to it I think it's better to base EQ on actual measurements and researched target curves, but again if that doesn't sound good to you then use something else....I did suggest using Oratory1990's raw measurements and then using REW to do your own EQ to different target curves of your choosing. But yeah, lets leave it, I've had enough of the discussion at this point. I'll continue to read around the subject once in a while to try to educate myself some more and to enable a deeper understanding, but I've definitely had enough of it for now....and for now I have the conclusions that I've been talking about in my last post.Its not unnatural if I perceive it as being tonally balanced with both noise and music. In the end, that's also what Harman did. Their curves could only be validated by actually letting subjects listen to that curves and adjusting the faders accordingly. Measurements and target curves without correlation to (subjective) human hearing and practical proving are of no real life use.
Yes, because it has little to do with the topic of subjective headphone equalization.
I suggest we totally skip on anything that includes the words "equal", "loudness" or "contour" and just talk about equalizing a headphone with a given set of bands to your liking until you subjectively perceive it as flat through the whole frequency band. That is what I was talking about all the time. I don't try to force any direct correlation to speakers, nor do I apply research that has been founded on listening to speakers in a (wet or dry) room. Listening to headphones which places the transducers right infront of your pinna is a totally differnt thing. There is no common standard for equalizing that particular reference point since it is not an established production standard. Hence, there is no right or wrong, only taste and subjective evaluation. And that is where the equalizer comes in which addressed a) irritating resonances which do not match your personal hearing and b) restores the tonal balance for the bass, mids and highs.
I recommend to start with the works by David Griesinger. He wrote some interesting papers about the correlation between loudspeakers and headphones, also putting head-related transfer fuctions and loudness perception into perspective.
He even has some of his presentations on YT:
(You can skip the proximity part if you like.)
He also explains why standard headphone equalizations made for the masses are mostly flawed and have to be corrected by the individual to achieve a balanced tonal presentation. That does not only apply to headphone-to-speaker calibrations (transfering frontal HRTF maps to speakers).
He even has some of his presentations on YT:
He also explains why standard headphone equalizations made for the masses are mostly flawed and have to be corrected by the individual to achieve a balanced tonal presentation. That does not only apply to headphone-to-speaker calibrations (transfering frontal HRTF maps to speakers).
Last edited:
OP
Regarding equal loudness eq I have a question. When we hear a flat speaker's sound, the sound will not be flat at our ear drum, but we think it is flat. When we do equal loudness matching for whole frequency response, is the result we ended up the same as flat speaker's sound at our ear drum? Or is it the flat speaker's sound outside of our ear, not at the ear drum?
From what I know the Harman preference curve is trying to copy the frequency response of flat speaker in a studio room, this also has the advantage of hearing what the producer is hearing in his studio. But there are many people saying the Harman preference curve is not flat, I can only come up with the reasons as follow. 1 the curve got modified a bit to correlate better with mass preference. 2 studio room is just one condtition, it does not mean flat speaker in studio room is flat. 3 the curve is flat, but because people got different ears, so flat for that person is not flat for everyone. 4 the curve is flat, but some people prefer not flat response and think it is flat. From those four reasons and your own reasons if you have, which ones are the cause? Also if you can, can you plot your own equal loudness eq and Harman preference curve eq on the same plot? I want to see how much is the difference and why people think Harman cruve is colored.
From what I know the Harman preference curve is trying to copy the frequency response of flat speaker in a studio room, this also has the advantage of hearing what the producer is hearing in his studio. But there are many people saying the Harman preference curve is not flat, I can only come up with the reasons as follow. 1 the curve got modified a bit to correlate better with mass preference. 2 studio room is just one condtition, it does not mean flat speaker in studio room is flat. 3 the curve is flat, but because people got different ears, so flat for that person is not flat for everyone. 4 the curve is flat, but some people prefer not flat response and think it is flat. From those four reasons and your own reasons if you have, which ones are the cause? Also if you can, can you plot your own equal loudness eq and Harman preference curve eq on the same plot? I want to see how much is the difference and why people think Harman cruve is colored.
All four to a certain degree ---
1) As the data collected from more test listeners is analyzed the average preferred curve is adjusted.
2) Speaker is anechoic measured flat on-axis and the listening room has a listening test proven 1dB/octave tilt 20-20kHz house curve at the dummy head measurement position.
3) The listening room reference curve is what the dummy head measures and listening tests adjust bass/treble to suit individual taste.
4) Range of bass/treble adjustment varies among groups - Experienced listeners tend to prefer the room reference curve measurement and there are variations based on age/sex/experience.
Depending on your personal taste you make EQ adjustments to the range of preferences to determine what sounds best to your own ear. If you're used to hearing transducers and rooms with uneven response (most of the general public) you may think none of the range sounds flat to you and reject it entirely. In that case, you need to spend a good deal of time listening to speakers engineered and designed with flat on-axis and smooth off-axis in a room that has a house curve similar to the proven preferred one so you know what to listen for in headphone response. I make A/B comparisons quite often between my room EQ'd studio monitors and personally adjusted reference Harman curve headphones for similar tonality.
Last edited:
OP
- Thread Starter
- #252
Well, I don't think that equal loudness matching EQ will result in a flat speaker or headphone full stop, so I don't really see it as a valid method, but that's just my opinion/take/analysis on it and others seem to disagree which is fine & I'm not about to start taking that up with some of them right now. Either way, it's main aim would be to try to create a flat sounding speaker or headphone, and as you probably know from the Harman Curve targets that this is not a flat target curve as measured at the ear drum on dummy heads, even though it's supposed to/does sound flat. E.g. You can see from headphone measurements & target curves that flat sounding does not equate to a flat frequency response when measured at the ear (https://www.dropbox.com/s/dm0m6u3s3b4zqzl/Sennheiser HD600.pdf?dl=0 ) So hopefully that answers some of your questions. (I do agree with the Harman Curve for headphones though, that's my personal recommendation).
OP
- Thread Starter
- #253
I updated the EQ on my headphones today (AKG K702), so thought I'd give a little rundown on what I've done & how.
My Previous EQ:
My previous EQ & how I arrived at it is described at my post here: https://www.audiosciencereview.com/...s-eqing-headphones-worth-it.11523/post-334811
To summarise how I did my EQ before (described in more detail at that link) it was based on poaching the EQ profile (by Oratory1990) from an Anniversary Edition of this headphone that is slightly different to the normal K702 that I own which is the bog standard version, and based on comparisons with Innerfidelity measurements of both K702 standard edition & anniversary edition I noted the main characteristic differences between the two and roughly adapted the Oratory1990 Anniversary Edition EQ profile by eyeballing it - I had to do it this way because Oratory1990 doesn't have an EQ for the K702 standard edition and I didn't trust the absolute accuracy of Innerfidelity measurements. So the following was the EQ I ended up with, my previous EQ:
My New EQ from Today:
So this part here is a rundown of my new EQ, and how I created it. I decided that my eyeballing EQ approach from before was not accurate enough, and also from comparing it with my Harman Curve roomEQ'd JBL 308p Mkii speakers it's clear that my previous EQ was lacking some treble energy/tone. Since my previous EQ on my headphones I had learned a lot in REW and acquired those JBL 308 speakers (hadn't even heard of REW before!), and decided to use REW combined with the actual RAW frequency measurements from both Oratory1990 and Innerfidelity to combine them....so essentially the same approach as before but not by eyeballing, more accurately this time. So I used 1kHz as the cut off point, combining Oratory1990 Anniversary Edition bass with Innerfidelity Treble. I didn't want to use K702 Innerfidelity Bass as there were some anomalies in there that I thought unreliable. I combined the two into a text file just by copying & pasting the data into a new text file, and another reason I chose 1kHz as the cut off point was because they're both 'normalised' at 1kHz with both being 0dB at that point from the measurements. I imported the file into REW so I could create manual EQ filters to the Harman Headphone Curve, here's a pic of the result in REW, manually created filters by me:
And this is what it looks like in PEACE:
You can see I also added a High Pass filter to start removing Low Shelf boost below 20Hz, as I've found it enhances clarity in the rest of the range (found that on my NAD HP50 headphones too). By looking at the PEACE curves you can also see that it's not massively far off my previous eyeballed EQ, but you can also see it does contain a little more treble energy. After listening today this sounds remarkably similar to my JBL 308 speakers, so I think I'm a lot closer than before with this headphone EQ, sounds great! I might still send these K702's to Oratory1990 to be properly measured & EQ'd and it will be interesting to see how far off my own EQ is from his. (previously I thought he lived in The States so didn't send them, but since found he lives in Austria!)
My Previous EQ:
My previous EQ & how I arrived at it is described at my post here: https://www.audiosciencereview.com/...s-eqing-headphones-worth-it.11523/post-334811
To summarise how I did my EQ before (described in more detail at that link) it was based on poaching the EQ profile (by Oratory1990) from an Anniversary Edition of this headphone that is slightly different to the normal K702 that I own which is the bog standard version, and based on comparisons with Innerfidelity measurements of both K702 standard edition & anniversary edition I noted the main characteristic differences between the two and roughly adapted the Oratory1990 Anniversary Edition EQ profile by eyeballing it - I had to do it this way because Oratory1990 doesn't have an EQ for the K702 standard edition and I didn't trust the absolute accuracy of Innerfidelity measurements. So the following was the EQ I ended up with, my previous EQ:
My New EQ from Today:
So this part here is a rundown of my new EQ, and how I created it. I decided that my eyeballing EQ approach from before was not accurate enough, and also from comparing it with my Harman Curve roomEQ'd JBL 308p Mkii speakers it's clear that my previous EQ was lacking some treble energy/tone. Since my previous EQ on my headphones I had learned a lot in REW and acquired those JBL 308 speakers (hadn't even heard of REW before!), and decided to use REW combined with the actual RAW frequency measurements from both Oratory1990 and Innerfidelity to combine them....so essentially the same approach as before but not by eyeballing, more accurately this time. So I used 1kHz as the cut off point, combining Oratory1990 Anniversary Edition bass with Innerfidelity Treble. I didn't want to use K702 Innerfidelity Bass as there were some anomalies in there that I thought unreliable. I combined the two into a text file just by copying & pasting the data into a new text file, and another reason I chose 1kHz as the cut off point was because they're both 'normalised' at 1kHz with both being 0dB at that point from the measurements. I imported the file into REW so I could create manual EQ filters to the Harman Headphone Curve, here's a pic of the result in REW, manually created filters by me:
And this is what it looks like in PEACE:
You can see I also added a High Pass filter to start removing Low Shelf boost below 20Hz, as I've found it enhances clarity in the rest of the range (found that on my NAD HP50 headphones too). By looking at the PEACE curves you can also see that it's not massively far off my previous eyeballed EQ, but you can also see it does contain a little more treble energy. After listening today this sounds remarkably similar to my JBL 308 speakers, so I think I'm a lot closer than before with this headphone EQ, sounds great! I might still send these K702's to Oratory1990 to be properly measured & EQ'd and it will be interesting to see how far off my own EQ is from his. (previously I thought he lived in The States so didn't send them, but since found he lives in Austria!)
Last edited:
- Joined
- Oct 25, 2019
- Messages
- 11,094
- Likes
- 14,751
You have a far more rigourous approach than I do, which is to crib the really wayward elements of the FR from oratory or wherever measurements and try and smooth them a little. More rounding the rough edges than aiming for a target. So my hd800 eq just has a W shaped dip at 6 and 9k and letting the rest stay as stock. I hate double digit preamp reductions too, means I have to re learn my amp volume positions.
A flat set of speakers is commonly linked to the response you can measure with a microphone at the tip of the triangle or the spot where the head would be. So as an absolute level, the response is always flat.
When the sound passes our head, pinna and ear canal the response will start to alter.
This response will change with the position of the sound source. Our brain tries to interpret those curves as "spectral maps" and compares them to our auditory experience we have grown up with. If the sound comming in matches those maps, the sound will be perceived as neutral. Deviations may be okay within a dregree because our brain is capable of adapting to different curves as long as they stay persistent.
Equal loudness matching - which transfers the HRTF response with a speaker infront of the listener to a given headphone - tries to recreate the exact pressure response at the eardrum. This can also be achieved with tiny microphones or sound tubes being inserted into the ear. We reference this as the response at the "drum reference point" (DRP).
Perceptively it can only be flat for those people whose spectral maps referenced at the DRP match the measurements and adjustments made by Harman or the specific headphone manufacturer. Since every ear is different there is great potential for not perceiving this curve as completely flat.
This interview with Paul Barton provides good explinations where those deviations come from.
As already emphasized in the research by David Griesinger, the exact length and shape of the ear canal has a stron effect on the timbre being perceived by the listener.
In practice there is no such thing as a perfect target curve, just some rough approximations.
Adjustments being derived from dummy heads will loose a lot of accuracy the higher you go in frequency because of their couplers not being able to emulate the actual high frequency response of the listeners ears. Notice that the old 60318-4 standard is only accepted to be accurate within 100 Hz -10 kHz. Some say that it is only valid until 8 kHz practically. The newer high-res couplers try to fix that with some sophisticated damping. Still, the whole subject is cursed as long as everyones ears are so different. The best you can get is approximations.
I tried Oratory's settings on my DT 880. But I couldn't get anything useful out of it above 6 kHz. It is at least 6 dB off and doesn't fix my resonance with the Beyer between 7 kHz and 11 kHz. Still not a bad starting point imo.
OP
- Thread Starter
- #256
Although I wouldn't have done all that if he already had a measurement for K702 'Standard Edition', he only has the Anniversary Edition, so it was about me using REW to blend two sets of RAW frequency data together and then me creating my own filters to fit the Harman Curve.....so it was the most accurate way I could 'bodge' it. It sounds very close to my Harman Curve EQ'd JBL 308's (eq'd using REW & UMIK), so I think that's a good reference to compare against. (My Sennheiser HD600 headphones arrived just now from UPS, by chance, I'm sure I'll post about them somewhere).
solderdude
Grand Contributor
- Joined
- Oct 25, 2019
- Messages
- 11,094
- Likes
- 14,751
Suspect using the 702AE measures would have been close enough.
OP
- Thread Starter
- #259
No, not really, there's a big dip in the treble in the K702 Anniversary Edition, that's not there in the K702, if you take a look at one of my old posts you can see by the graphs (at 3000Hz):
https://www.audiosciencereview.com/...s-eqing-headphones-worth-it.11523/post-334811
- Joined
- Oct 25, 2019
- Messages
- 11,094
- Likes
- 14,751
I get that. I just wonder in real world listening how noticeable that would be
Similar threads
