[Audman71]

I like to experiment with audio, but one of my favorite things to do is make headphones I have that are mediocre quality to be superb in terms of frequency response. The equalizer is my go-to tool for the job. Recently I started working with my stock apple EarPods. The way I went about it, I was mostly guessing, and not using any sort of hardware equipment or other tools that may be used for this sort of testing, as I do not own such equipment nor do I currently know how to use it. As such, there may be some (or many) things technically incorrect with what I’ve done. Taking that into account, this is the general process I went through to get the results I have as of today:
 
First, I used ReaFir with an FFT size of 32768 to create this eq pattern:
 

 
 
It is an attempt at an approximate inverted version of the following graph:
 

 
 
It sounds about what I hoped would happen; a massive improvement, by my standards.
 
Afterwards, I decided to make further adjustments, but this time I tried fine tune two other instances of the equalizer for my left and right ears separately, to get the 20-20000hz sine sweep sample I got online to sound as close as I can get to even to each of my ears.
 
This is my right:
 

 
 
And my left:
 

 
 
The result under the combined settings of my attempt at EQ correction of my Apple EarPods and each of my ears was a much darker sound to audio I played back than I had previously heard. As a side note, I thought it sounded generally better at higher volumes than both the un-EQ’d audio and the EQ’d setting I had before I tried to fit the sound to my ears.
 
Here’s why I bring this topic up: I don’t know if I know enough about what I did to make a proper judgement call on whether or not I accomplished anything with these settings, or if I just created a really good sounding and complex way to make a setting on an equalizer.
 
What do you guys think? What exactly did I do here?

 

[castleofargh]

the first work with goldenears graph is you trying to get the sound they decided was neutral. you could like it, or not, and it could sound neutral to you, or not. still I would expect it to sound pretty fine overall given their target response.
little advice though, you should not try to compensate the roll off at both ends of the frequency response like you did. if the IEM rolls off in the bass and trebles, it's most likely because of physical difficulty. by pushing harder on those difficult frequencies, you're most likely creating a lot of distortions without really bringing the sound where you expect it to go. for the bass I would stop trying to rise at where the 50khz is. but that's just me. and same for trebles.
second warning, when you read IEM or headphone measurements, the trebles area has no guaranty to be what the graph shows. both because of how it's measured, and because the way you put the IEM in your ears and how your ears are, both will affect the signature a great deal in the trebles. so after maybe 10khz, you should take the measurements with a grain of salt and really trust your ears more than any graph.
 
 
the second part of your job with a sine wave, if you tried to make all frequencies sound as loud, then supposedly you created an equal loudness contour for your ears with that IEM. the "problem" in a way is that your EQ doesn't really look like what I would expect from an equal loudness contour. did you do it while the first EQ was ON?
a real equal loudness contour, if that's what you achieved, would not sound neutral at all as it's something we live with all our life. it's a default of our hearing, but it's the only way we know of listening so compensating for it can only sound weird.

 

[watchnerd]

I'd like to see some distortion measurements.
 
 
Anyone can use an EQ to smash a transducer into flatness. Whether that's smart and what that does to driver distortion is a different issue.

 

[Audman71]

  the first work with goldenears graph is you trying to get the sound they decided was neutral. you could like it, or not, and it could sound neutral to you, or not. still I would expect it to sound pretty fine overall given their target response.
little advice though, you should not try to compensate the roll off at both ends of the frequency response like you did. if the IEM rolls off in the bass and trebles, it's most likely because of physical difficulty. by pushing harder on those difficult frequencies, you're most likely creating a lot of distortions without really bringing the sound where you expect it to go. for the bass I would stop trying to rise at where the 50khz is. but that's just me. and same for trebles.
second warning, when you read IEM or headphone measurements, the trebles area has no guaranty to be what the graph shows. both because of how it's measured, and because the way you put the IEM in your ears and how your ears are, both will affect the signature a great deal in the trebles. so after maybe 10khz, you should take the measurements with a grain of salt and really trust your ears more than any graph.
 
 
the second part of your job with a sine wave, if you tried to make all frequencies sound as loud, then supposedly you created an equal loudness contour for your ears with that IEM. the "problem" in a way is that your EQ doesn't really look like what I would expect from an equal loudness contour. did you do it while the first EQ was ON?
a real equal loudness contour, if that's what you achieved, would not sound neutral at all as it's something we live with all our life. it's a default of our hearing, but it's the only way we know of listening so compensating for it can only sound weird.

 
I should have mentioned earlier, I was using Reaper and not just the plugin ReaFir. I compensated for any distortion I noticed by turning down the level in Reaper's main window and not in the plugin. I turned it down to what turned out to be -13 db. It's quieter in Reaper, but if I turn up Windows's volume control, it can sound about the same in terms of loudness. If you're talking about headphone or amp-specific distortion, I have never noticed any distortion with my computer's built-in sound card with the apple earbuds. Sound artifacts, maybe, but not distortion.
 
As for your question, yes, As I said, I did equalize the ear EQ's by combining the settings of the three equalizers, meaning all three equalizers were on. I just panned the left and right ear EQ's to the left and right speakers, respectively.

 

[Maconi]

The EQ program you're using looks decent (I've been using Equalizer APO + Peace GUI but have been looking at other options).
 
If you're boosting frequencies make sure to use the Preamp feature to keep the highest boost @ 0dB (so everything else might be -30dB or whatever), which it looks like you're already doing.
 
Also max out the volume everywhere and cut the volume using only the Preamp (or your external amp) as cutting the volume in Windows or some applications will lose audio bits and destroy the audio quality.
 
Also, I wouldn't bother with trying to match it perfectly with graphs, every ear is different (even if the headphone measures perfectly to a machine, it may sound peaky to your ear). Use a Sine Generator and have it sweep from 15Hz up to 20kHz and use EQ to make every frequency sound equal (which once again it looks like you're already doing).
 
As mentioned above though, some headphones just can't handle being forced flat. Your best bet is trying to find a headphone with the flattest response and the best THD you can afford (the worse the THD the more easily it will distort with EQ). It's also easier to cut than boost (so avoid headphones with massive dips, spikes can be okay).
 

 

 

[Audman71]

 
As mentioned above though, some headphones just can't handle being forced flat.

 
I don't know about that. Maybe its just me, but the Apple EarPod earbuds I used for this test seem to sound a lot better with no distortion that I can hear after I messed with the EQ. The only distortion I ever heard from a signal that was too loud came from my software's preamp, not the headphones. Considering the initial quality of the headphones, maybe I'm not focusing on the right thing?
 
castleofargh: I have since EQ'd the headphones to compensate for a tight fit (following a different graph from the same site). It sounds less dark and better than what I had before. The bass increase reaches 0 db at 35 hz now instead of 50 hz, and the treble's have been slightly altered.
 
Also, just in case it was unclear, When I did this EQing process, I started by trying to flatten the EQ for the headphones by inverting the graph in my first post. Then, leaving the first equalizer instance running with a theoretically flat EQ as the base, I adjusted two additional instances of the same equalizer, one panned left and one panned right, to fine tune the signal to be the closest to flat that I could for my ears.

 

[watchnerd]

   
I don't know about that. Maybe its just me, but the Apple EarPod earbuds I used for this test seem to sound a lot better with no distortion that I can hear after I messed with the EQ. The only distortion I ever heard from a signal that was too loud came from my software's preamp, not the headphones. Considering the initial quality of the headphones, maybe I'm not focusing on the right thing?

 
Points:
 
1. Your ear is less sensitive to distortion of bass, especially sub-bass frequencies, than at other frequencies.
 
2. What signal did you use to test it?
 
3. What SPL did you test at?

 

[Audman71]

   
Points:
 
1. Your ear is less sensitive to distortion of bass, especially sub-bass frequencies, than at other frequencies.
 
2. What signal did you use to test it?
 
3. What SPL did you test at?

The only question I can answer is #2, as I don't actually have any special equipment for audio testing, and the volume level I listened to the signal at may have varied. I don't remember. As such, the "signal" I used to test it was a logarithmic sine sweep audio file from http://www.audiocheck.net/ and not a signal generated by hardware or software.

 

[castleofargh]

 

 
As mentioned above though, some headphones just can't handle being forced flat.

 
I don't know about that. Maybe its just me, but the Apple EarPod earbuds I used for this test seem to sound a lot better with no distortion that I can hear after I messed with the EQ. The only distortion I ever heard from a signal that was too loud came from my software's preamp, not the headphones. Considering the initial quality of the headphones, maybe I'm not focusing on the right thing?
 
castleofargh: I have since EQ'd the headphones to compensate for a tight fit (following a different graph from the same site). It sounds less dark and better than what I had before. The bass increase reaches 0 db at 35 hz now instead of 50 hz, and the treble's have been slightly altered.
 
Also, just in case it was unclear, When I did this EQing process, I started by trying to flatten the EQ for the headphones by inverting the graph in my first post. Then, leaving the first equalizer instance running with a theoretically flat EQ as the base, I adjusted two additional instances of the same equalizer, one panned left and one panned right, to fine tune the signal to be the closest to flat that I could for my ears.

yup that's what I figured from your previous answer. but I still don't know if you're going for equal loudness or for flat to your ear.  it's the same doubt I have with http://www.head-fi.org/t/796791/the-most-reliable-easiest-way-to-eq-headphones-properly-to-achieve-the-most-ideal-sound-for-non-professionals
I fail to understand how listening for even loudness of frequencies can lead us toward anything other than an equal loudness contour curve.

 

[watchnerd]

 
yup that's what I figured from your previous answer. but I still don't know if you're going for equal loudness or for flat to your ear.  it's the same doubt I have with http://www.head-fi.org/t/796791/the-most-reliable-easiest-way-to-eq-headphones-properly-to-achieve-the-most-ideal-sound-for-non-professionals
I fail to understand how listening for even loudness of frequencies can lead us toward anything other than an equal loudness contour curve.

 
The Fletcher-Munson curve is dynamic.
 
At best, you could *try* to equalize to it as a target at a specific SPL, but as soon as you go above or below that your EQ is wrong.

 

[Audman71]

   
The Fletcher-Munson curve is dynamic.
 
At best, you could *try* to equalize to it as a target at a specific SPL, but as soon as you go above or below that your EQ is wrong.

 
Actually, through experimenting with the software that came with the ReaFir plugin you see in the picture, I figured out that another plugin by the same company, ReaEQ, allows you to change your EQ curve dynamically by setting a few parameters. It works relatively the same, but is less flexible in certain ways, which is why I used ReaFir, although ReFir to my knowledge does not have the ability to be dynamic.

 

[watchnerd]

   
Actually, through experimenting with the software that came with the ReaFir plugin you see in the picture, I figured out that another plugin by the same company, ReaEQ, allows you to change your EQ curve dynamically by setting a few parameters. It works relatively the same, but is less flexible in certain ways, which is why I used ReaFir, although ReFir to my knowledge does not have the ability to be dynamic.

 
Well, you also have the problem of the mix.
 
Namely, at what volume was the mixer listening when he made the mix?
 
Most guys I know balance the mix so it sounds 'right' at a moderately loud level.
 
Below that, Fletch Munson kicks in...
 
...then again this is just the old 'loudness' button all over again.

 

[gregorio]

  Use a Sine Generator and have it sweep from 15Hz up to 20kHz and use EQ to make every frequency sound equal (which once again it looks like you're already doing).

 
Unless you are only listening to tones from a sine generator, why would you want to compensate for equal-loudness twice?
 
G

 

[Audman71]

   
Well, you also have the problem of the mix.
 
Namely, at what volume was the mixer listening when he made the mix?
 
Most guys I know balance the mix so it sounds 'right' at a moderately loud level.
 
Below that, Fletch Munson kicks in...

Well, there's nothing that can really be done about the mix aspect of it. Everyone's ears are different anyway. Not to mention the equipment used, the songs themselves and their volume, effects, etc.
 
I looked up that curve you were talking about. As what I am doing is a work in progress, I have decided to follow pictures, add the Fletcher Munson Curve (roughly at the 60 dB measurement) as best I could to an instance of ReaFir, and redo customizing the EQ to fit my ears with both the inverted headphone graph and the Fletcher Munson Curve equalizer setting active at the same time.
 
It got much brighter, better sounding than my original result, in my opinion. After testing my settings with a sine sweep (20-20000hz), it sounds pretty much even to my ears. Even on the small and cheap Apple EarPods, it sounds good. There is a small amount of buzzing, and I may need to get another pair for testing purposes, though. Those of you who thought some headphones can't handle being flattened out seem to be correct.
 
Aside from that and the frequencies generally sounding good, is it a statistical improvement?
 
EDIT:
 
 

 
I still don't know if you're going for equal loudness or for flat to your ear.

 
I'm sort of going for both, if that makes any sense. I want it to sound as flat as possible and still sound good. The Fletcher Munson Curve, from what I gather, while not purely even in frequencies, is an equal loudness curve tuned for the average human ear. I wanted to make the headphones sound as close to perfectly flat to my ears as I could, so I tuned the left and right ears in the equalizer with the sine sweep mentioned above after applying the Fletcher Munson Curve to the signal. But before I could do any of that, I had to make the headphones theoretically flat according to a graph so I could get even close to the proper measurements.
 
In a nutshell, the signal flow I currently have is the following, chronologically:
 
1: Audio Signal
2: Approx. Inverted Apple Earpod EQ From Graph
3: Approx. Fletcher Munson Curve Applied Via Equalizer
     4a: Fine-Tuned EQ For Ears (L)
     4b: Fine-Tuned EQ For Ears (R)
5: Apple EarPods
 
Hope this helps.

 

[gregorio]

Originally Posted by Audman71 /img/forum/go_quote.gif
 
I have decided to follow pictures, add the Fletcher Munson Curve (roughly at the 60 dB measurement) as best I could to an instance of ReaFir, and redo customizing the EQ to fit my ears with both the inverted headphone graph and the Fletcher Munson Curve equalizer setting active at the same time.

 
As I asked Maconi, why would you want to compensate for the FM curve twice?
 
G