What do people get when they build curves with this tool:
Does anybody else's really closely resemble the established curve? I find mine to be a little different (though it's sorta hard to tell without some curve fitting)
I did twice, once with my shure 1840 and another with my HE500, I seem to measure flat from 500 to 4000 Hz with the 1840 and flat from 375 to 6000 Hz with the HE500. I don't seem to have that hump at 1.5 kHz but everything else seems to be good. Pretty cool experiment.
Yeah, I got started thinking about this when I saw the etymotic frequency plot at the bottom of the page (http://www.etymotic.com/ephp/er4.html).
I figured well, if they claim that the perceived response curve was really that flat, couldn't I test it some way? So I found that tool, and with ER-4P on my Macbook Pro, there was a noticeable curveature...
30Hz = -18dB
45Hz = -20dB
60Hz = -24dB
90Hz = -33dB
125Hz = -36dB
187Hz = -39dB
250Hz = -42dB
375Hz = -45dB
500Hz = -48dB
1kHz = -48dB
1.5kHz = -54dB
2kHz = -57dB
3kHz = -63dB
4kHz = -69dB
6kHz = -75dB
8kHz = -69dB
12kHz = -48dB
16kHz = -27dB
It's sorta hard to tell by looking at the numbers only, but the whole thing is curved.... now of course, I'm not exactly using the highest fidelity system, but I guess it sorta stands to reason:
1) How much of this is due to my ears
2) How much of this is due to my setup?
Quaintative, if your headphone chain is relatively flat, your response curve should look like the already made equal loudness contours. I don't know how Etymotic got that second graph to be the 'perceived frequency response' of their earphones, but your results show that they are rather close to neutral-- although seemingly high in energy around the 6k area.
I got something like this using Magni/Modi > HE-400:
This test was hard to take for me, it's hard to be very accurate about things and judging overall sound perceived audio energy for each band. The only thing that sticks out with the HE-400 is the 12khz band being lower than the general contour, really representational of the HE-400's massive treble spike in the 12khz area.
This is what mine looks like:
Well, I would presume that Etymotic simply did a subtraction of their curve from the equal loudness curve. On a log scale, these things might look reasonably "similar". I was more surprised at how much a deviation from the "perceived" curve it looked. I could use the P-to-S adapter and see if that flattens it out, since it tends to strengthen the higher frequencies, thus reducing the dip I see at 6 kHz.
It's just one of those things.... are headphones more precise? or are our ears simply a little bit more varied than we think (and is it worth EQing out?)?
Strengthening the higher frequencies would only make the dip at 6khz even lower-- if you consider 6khz a highish frequency anyways.
I don't really think our ears are all that different as much as some people on head-fi pretend to. I do know that hearing loss past 15khz from age and exposure to loud noise is very real though.
Oops, yeah, you're right. If I was going to try to use this as an EQ implementation, I'd have to flip it and then use only the relative dB differences.
And sure, our ears may be pretty similar along with high frequency degradation. But it sure doesn't hurt to try something to try to flatten out the frequencies. I figure with an even frequency response, it'd be easier to tell the differences in other characteristics of the headphones (ie. response time, detail/clarity, etc.).
Yes, theoretically you would flip it and find the difference between your personal equal loudness contour and the provided universal equal loudness curve, although it would still be difficult because the universal equal loudness curves don't use the same y axis, and I wouldn't know how to align the two different curves up perfectly either.
It can be a good EQ tool to find things that stick out like a sore thumb though, for instance the 12khz on the HE-400 in my instance and the 6khz on your Etymotic.
Explain? Sines and music frequencies aren't that far apart, given the decoding algorithm. I was under the impression that waveforms were decoded with an assumption to look like sine waves.
And yes, how the software decides to deal with the intervening frequencies between say... 1kHz and 1.5kHz could introduce more frequency distortions then fix them (resulting in a worst case scenario of two areas of distortion per frequency range), but given a good enough curve, couldn't you standardize frequency curves across headphones.
Of course, if you wanted to be super particular about this, you'd have to consider a non-linear response by the microphones and pickups in the first place.... but focusing on the headphone output only....
The equal loudness curves do use the same y axis, each curve simply refers to a different sound pressure at 1kHz (I think). You could still use the loudness curve as reference values in relation to each curve shape.
Pure tones and music are quite a bit apart, but that doesn't matter primarily here. We measure loudness in phons, but sound pressure in dB SPL. When you look at measurements of calibrated speakers you will see a flat line and the y-axis labeled sound pressure [dB SPL]. On a loudness graph the calibrated speaker would look very close to the equal loudness contours, i.e. not flat at all.
Consider going to a live performance. If during this performance, tones where reproduced for the audience, the different frequencies will be perceived with different loudness levels (as illustrated by the loudness contours.) Does that mean that one needs to somehow equalize a live performance, or for that matter the entire world, to "correct" for loudness levels? I don't believe so.
The job of a high fidelity audio reproduction system is to reproduce a live performance, or a particular set of effects, as close as possible to the original source. This implies that the lumped sound system frequency response should be as flat as possible so as to "get out of the way."
If equalizing using the loudness contours, what will happen is that a violin or any other instrument played through the equalized rig will sound substantially different than a "real world" violin or any other instrument.
Let's take any instrument as sound source. We record this with a perfect mic (completely flat frequency response). We replace the instrument with a perfect speaker and play the recording. We record this reproduction with the perfect mic again and get exactly the same as the initial recording.
So what's the problem? The problem is your torso, head, outer ear, ear canal ... All of this changes the frequency response of what actually arrives at your eardrums, aka head related transfer function (HRTF). Different types of equalization (diffuse, free, independent-of-direction) try to account for that by reversing the effects.
Ideally, you'd measure the instrument or perfect speaker with a perfect mic at your eardrum and compare it to the measurement of your headphones. You'll see a resonance around 3 kHz etc. That is what you need to equalize for, not equal loudness of pure tones.
Correct. You absolutely DO NOT want to equalize out your equal loudness curve. That curve set is built into human hearing, you don't want to equalize it out, and besides, it's level dependent.
The curve you'll get from this test will be the equal-loudness response of your own hearing with the mask of the response of the headphones/IEMs you used for the test. The results are pretty meaningless and not useful at all for setting an equalizer. At best the results present something that is of interest to look at.
The target response curves of the three types of headphones plus IEMs are all different, but none are flat, and certainly none are related to the equal-loudness curves. Headphone target curves are there to compensate for the difference in the way sound is presented to the hear by headphones vs a free field.
You can experience highly accurate and detailed headphone EQ with the Audyssey "amp" app, and some of the others. Audyssey also features their dynamic EQ, which compensates for the effect of the change in hearing response at low levels relative to the spectral balance at the level at which music is mixed.
Using an equalizer to compensate for an individual's hearing difference is also not a great idea because every individual has a build-in established "normal" based on every-day hearing. Attempting to compensate for a deficiency, unless large enough to be considered a hearing loss, will always result in unnatural balance because of its contrast with what's heard in life. It's for this reason that people who have hearing aids prescribed are encouraged to use them all day, every day, or they will always sound "wrong".