[Joe Bloggs]

Here's me running two EQs in series: the equal loudness compensation curve on right, my actual settings for the Philips SHE3580 on left.  The compensation curve is taken out when playing music.
 

 
On the right I have Sinegen playing the different peaks I have found, piped through Virtual Audio Cable to the two EQs in VSTHost before outputting to headphones.  With the equal loudness curve accounted for I just need to concentrate on making everything sound as loud as 60dB.  Never tried the pink noise thing, don't know how that's supposed to work.
 
Not saying my method is perfect (bass sounds too thin if I try to EQ it flat with this config--but I just tried to administer this test to my brother with the same phones and he actually EQed the bass up and then it sounded too thick to his ears.  So there's tester variance here) but it should be closer to the truth.

 

[Argyris]

Never tried the pink noise thing, don't know how that's supposed to work.

I didn't get it at first, either. It's more of a gut feeling than an absolute. You find your peak with the sine sweeps, set up a notch over the top of it in your EQ, play the pink noise sample, and drag the center of the notch down until the peak "disappears" from the noise. I'm not sure how to describe what exactly that sounds like--I just know when it's right. It takes a few times because I feel differently about where to stop depending on the iteration--I usually just remember which values kept coming up and use those as an average.


[quote="Joe]

Not saying my method is perfect (bass sounds too thin if I try to EQ it flat with this config--but I just tried to administer this test to my brother with the same phones and he actually EQed the bass up and then it sounded too thick to his ears.  So there's tester variance here) but it should be closer to the truth.
[/quote]


There absolutely should be a difference between individuals, even leaving out personal preference. The equal loudness curves are aggregates of a lot of individual tests, and as such they probably don't exactly correspond to any living (or destined to be living) person's actual curve (though they're probably close enough to make a good benchmark). Also, they'll differ depending on what amplitude the reference tone (1kHz, I think they use) starts at. That means that, assuming the standard curve fits your hearing closely enough, the amplitude you use to test needs to be calibrated to that used to generate the specific standard curve you've modeling.

Hopefully that made some sense. If not, I could try again because this stuff is really confusing to try to describe.

 

[nivrethejugg]

Equalizing has never been something that I have never known how to do properly and this guide showed me the proper way I should go about doing it so I would like to say thanks to the OP. It has come a point where I can barely stand music not equalized with the ER4S on. Going back and forth definitely made me want to stick with it being equalized and more neutral.
 

 
The spikes above kind of shows how the ER4S can be harsh and I found that without reference to this chart, I have similar results. Loudness at 3kHz and 11kHz. So I tamed them down and the ER4S is neutral. Results may be because of my sound card so I wouldn't be too surprised if it differed between sources.
 
 

 
Once again, thanks for the tutorial. Made me appreciate my ER4S a lot more. Now... I'm not sure if increasing the bass would be a good idea. I have heard people mention that the bass for it is weak. I think it sounds pretty good right now but let's see how well it'll go if I did increase the bass. 

 

[Argyris]

Now... I'm not sure if increasing the bass would be a good idea. I have heard people mention that the bass for it is weak. I think it sounds pretty good right now but let's see how well it'll go if I did increase the bass. 

EDIT: Forget what was here before if anybody already read it. I kind of got technical and speculative, so I'll just say generally that it's all about what you want to hear. If it sounds good, then don't let others' assessment alter that. However, 'tis no crime to add a bit of extra thump. I've got a slight, 3dB low shelf on my DT880 to add a bit of body (and to compensate for a slight loss due to the crossfeed I'm using). It's slight but it makes all the difference.

Let us know how it works out.

 

[nivrethejugg]

Alright, I have been fiddling around for a little while and I still feel that not boosting the bass is the better choice. The added colour makes me lose clarity but I must say if it were not for the loss of clarity, I would have done it. Maybe I should get some olives and re-equalize. Might sound better.

 

[OPrwtos]

deleted!
 
(old superfluous post)
 

 

[AstralStorm]

Quote:

@AstralStorm: How can plain EQing by ear give correct results when we perceive e.g. 30Hz to be 40dB quieter than 1kHz when they are in fact the same volume?  The result will be an EQ setting that way overemphasizes the bass and treble.  I bet you'd wanted to push those 20-30Hz sliders up past the ceiling but then decided it didn't sound good with actual music so dialed them back after trying to convince your brain that those levels are level with the 500Hz tone when they clearly weren't

 
I equalize by ear too.  I do that AFTER accounting for the equal loudness curve.  If I didn't I would just leave all the bass frequencies flat with 1kHz on my Philips SHE3580 EQ.  And hurt my brain with skull-crushing bass every time I play actual music

 
Also after I went to continuous tone sweeps with Sinegen and continuous FR adjustments with parametric EQs like Electri-Q I'm never going back to discrete tone measurements and graphic equalizers

Uhm... I equalize to percieve the exact same loudness for all tones. This takes the "equal loudness curve" into account naturally - no need to compensate for that at all.
The only important step for comparative results is to calibrate the reference level reasonably well. I pick ~40 sones, that is normal conversation loudness. (using a bunch of radio recordings)
 
Note the mastering engineer also cut to what he heard was right. You do not want to invert or double the curve - everyone's brain has some version of it built-in. It means that tones of given pressure level will sound exactly of same loudness. If they sound equally loud, then the frequency response is accurate.
 
Almost all headphones have problems reproducing lowest frequencies loud enough. That requires moving large amounts of air and thus strong magnetic field, large surface area, both very hard to achieve in headphones and in IEMs. IEMs have the plus of better seal (thus higher efficiency) and closer distance to eardrum.
Guess why subwoofers are large and use huge magnets...
 
Suprisingly, subbass equalization seems to work best with dynamic driver IEMs and not headphones. Even great headphones (like DT-1350, which are nigh impossible to overload according to InfinityAudio) start to show distortion then. (quite obvious "fart" rough bass)
BA IEMs also can't typically reproduce enough subbass and they respond even worse to subbass eq - perhaps with exception of multi-BA designs - Westone 4 was quite equalizable in the lowest end...
 
Of course IEM bass is less visceral than headphone bass, which is still less visceral than speaker bass - which is still typically inferior to the real thing.
 
For comparison, RE272 (allegedly "bass light" dynamic driver IEM - more suggestive of fit issues) does about equally as well on subbass front, but better in high freqs and is far more detailed in general. Note the reference is lower in volume, 20 dots on FiiO E7, as these are quite a bit more efficient and also low-impedance. (20 Ohm)
Note the boost ~20Hz would be about +16dB to be actually equal loudness, I didn't want to compromise other apps.
 
With the eq, it's equally loud 30Hz-17kHz. (But it's noticeable that it's extended to 20kHz or so, extension is not measurable by my ear, can detect up to "just" 19.5kHz.)
 
The ~40 Hz resonance is something specific to my own ears, bone conduction resonance?

Forgot to mention: Same FiiO E7 wideband 4.5 dB bass boost. That's the reason for the dip in there.
 

 

[Joe Bloggs]

Quote:

Uhm... I equalize to percieve the exact same loudness for all tones. This takes the "equal loudness curve" into account naturally - no need to compensate for that at all.
The only important step for comparative results is to calibrate the reference level reasonably well. I pick ~40 sones, that is normal conversation loudness. (using a bunch of radio recordings)
 
Note the mastering engineer also cut to what he heard was right. You do not want to invert or double the curve - everyone's brain has some version of it built-in. It means that tones of given pressure level will sound exactly of same loudness. If they sound equally loud, then the frequency response is accurate.

 
The mastering engineer cut to what he heard was right on a reference speaker system, which means equal actual response for all frequencies.  This will be perceived as light on bass when playing test tones at equal actual loudness, and when playing test tones only.  A system equalized to equal perceived loudness playing test tones will have a frequency response roughly similar to one of these curves

You said you calibrate to 40 phons so the 40 phon curve should be the closest.
 
Note the similarity in shape to your RE272 EQ curve

 
Of course human hearing is a funny thing and in time you may feel that such a response is the "right" response

 

[AstralStorm]

Actually, shape is not similar at all, the total bass difference is max +13.5 dB at 20 Hz, while equal loudness curve is +60 dB.
That's many orders of magnitude more. The shape is due to smooth subbass rolloff of the IEM.
 
Also, reference speakers are designed to have a flat perceived response. That's what is meant by high fidelity - sound recorded correctly will be perceived (as close to as possible to) identically as the live source. It so happens that with speakers and room design, that's achieved at flat frequency response.

 

[Joe Bloggs]

By definition for the sound recorded to be recorded correctly and played back "correctly" they have to be both recorded and played back with a flat actual frequency response, perceived FR does not come into any of this except for our perception of what sounds good, IOW why we prefer to listen to guitars and drums instead of white noise.  That's why if you look at the spectrum of just about any music recording you will find the energy is heavily slanted toward the bass end.  In other words the difference between perceived FR and actual FR is compensated for by the actual music (rather than test tones) having a mix of spectral content biased towards the bass.  Now if you made your system have a flat perceived response to test tones, where does that leave actual music?
 
Also,
Quote:

Also, reference speakers are designed to have a flat perceived response. That's what is meant by high fidelity - sound recorded correctly will be perceived (as close to as possible to) identically as the live source. It so happens that with speakers and room design, that's achieved at flat frequency response.

Are you suggesting that a pair of speakers with flat actual FR placed in a typical room would appear to play all 40dB test tones at equal perceived volume at 40 phons?  So a typical room's acoustics would bump 20Hz up by 60dB relative to 1kHz?

----
Since in these internet arguments we are really looking for somebody to agree with us while the other participant of the argument will be the last person on earth to ever agree with you, do you mind if I take a different approach:  you people reading this thread--who's with me?

  Who's with AstralStorm?

 

[SanjiWatsuki]

I agree with you. Especially for the lower bass, it's normally compensated for by the studio. 

This is the frequency response for the KRK K-Rok, an old school studio monitor. For the most part, it's neutral, except for the bass roll-off. If it's being heard in the studio using something like this, do you think that it would be played back correctly by having a flat bass response? Most likely not, you'd probably need bass roll-off to get the intended sound. 
 
Quote:

By definition for the sound recorded to be recorded correctly and played back "correctly" they have to be both recorded and played back with a flat actual frequency response, perceived FR does not come into any of this except for our perception of what sounds good, IOW why we prefer to listen to guitars and drums instead of white noise.  That's why if you look at the spectrum of just about any music recording you will find the energy is heavily slanted toward the bass end.  In other words the difference between perceived FR and actual FR is compensated for by the actual music (rather than test tones) having a mix of spectral content biased towards the bass.  Now if you made your system have a flat perceived response to test tones, where does that leave actual music?
 
Also,

Are you suggesting that a pair of speakers with flat actual FR placed in a typical room would appear to play all 40dB test tones at equal perceived volume at 40 phons?  So a typical room's acoustics would bump 20Hz up by 60dB relative to 1kHz?

----
Since in these internet arguments we are really looking for somebody to agree with us while the other participant of the argument will be the last person on earth to ever agree with you, do you mind if I take a different approach:  you people reading this thread--who's with me?

  Who's with AstralStorm?

 

 

[AstralStorm]

Quote:

Are you suggesting that a pair of speakers with flat actual FR placed in a typical room would appear to play all 40dB test tones at equal perceived volume at 40 phons?  So a typical room's acoustics would bump 20Hz up by 60dB relative to 1kHz?

 

Actually no. There are two schools out there, one that has the "Loudness correction" on, and the other that has "Loudness correction" off. Mastering is done typically without loudness correction, except for most of electronica.
 
The assumption of course is that music is typically composed of clear tones. This includes synth bass. The other kind of equalization would probably use bandlimited pink noise, simulating melodic drums. I've already mentioned that. There will be some difference of course, but not that much. Also remember, any sound can be decomposed into a sum of sines, per Fourier's transform.
 
The third impossibly expensive correction would use an SPL measurement at the eardrum in comparison with a reference loudspeaker. Good luck getting that kind of gear that can record very low frequencies well too.
 
All in all, it's whether what you hear matches the live performance.
 
Also watch out, there are other dB out there, esp. dBV (power) which is common in loudspeakers, and of course dB(A), the A-weighting equal loudness approximation.
I haven't yet seen an actual dB SPL measurement out there.
 
Here's for some differences:
http://www.headphone.com/learning-center/build-a-graph.php?graphID[0]=2859&graphID[1]=2861&graphID[2]=&graphID[3]=&graphType=0&buttonSelection=Compare+Headphones
Red (first) is Raw, Blue is Compensated. Note that HeadRoom doesn't bother to compensate bass, as that's really notoroius for being hard to measure and reproduce. Even fit can cause +/- 12 dB there.
 
Also, this is some good writeup on the differences: http://www.inearmatters.net/2008/12/neutral-vs-natural-thought.html

 

[Joe Bloggs]

 
Quote:

 
Also watch out, there are other dB out there, esp. dBV (power) which is common in loudspeakers, and of course dB(A), the A-weighting equal loudness approximation.
I haven't yet seen an actual dB SPL measurement out there.

 
Quote from wikipedia:
'Probably the most common usage of "decibels" in reference to sound loudness is dB SPL, sound pressure level referenced to the nominal threshold of human hearing:
 
dB (sound pressure level) – for sound in air and other gases, relative to 20 micropascals (μPa) = 2×10⁻⁵ Pa, the quietest sound a human can hear. This is roughly the sound of a mosquito flying 3 meters away. This is often abbreviated to just "dB", which gives some the erroneous notion that "dB" is an absolute unit by itself.'
 
So dB(SPL) is the assumed unit for any FR measurement where the unit is not specified or just in "dB".
 
So for example the FR of the KRK loudspeakers SanjiWatsuki just posted is in dB(SPL).  And it curves not up, but down in the bass frequencies.

 

[SanjiWatsuki]

My point is just that they don't shoot for complete neutrality into the sub-bass frequencies. To have an utterly flat bass FR like an LCD-2 is colored in its own way, I'd argue.
 
Quote:

 
So for example the FR of the KRK loudspeakers SanjiWatsuki just posted is in dB(SPL).  And it curves not up, but down in the bass frequencies.

 

 

[Joe Bloggs]

Yeah, since the vast majority of loudspeaker systems are also rolled off in the bass (insane subwoofer systems using a whole basement as a loudspeaker enclosure excepted), so they would try to make music sound good in spite of such a rolloff by boosting the bass in the mix itself.  And my other point was that even a system with flat measured bass response playing bass tones would sound much quieter than it playing a 1kHz tone at the same SPL because our ears' FR is also rolled off.  So if you equalize bass and treble test tones that are supposed to produce the same SPL to be the same loudness you would get doubly exaggerated bass response.