How far can EQ really go towards truly equalizing headphones?

[bigshot]

What about distortion? The reason speakers have so much of a problem reproducing square waves is because the push and pull is so instantaneous, the flapping paper cones can't react fast enough. This causes distortion, making the square wave not very square. The problem isn't the frequency of the square wave, it's the unnatural shape of it.

 

[xnor]

Quote:

Originally Posted by ultrabike /img/forum/go_quote.gif
 
With that said, a square wave is a relatively easy sound test for speakers and headphones IMHO.
[...]
It is therefore my opinion, that one cannot expect accurate reproduction of music if a simple 300 Hz sinusoidal along with it's first 4 odd harmonics cannot be accurately reproduced by it.

But it is not. Headphones and speakers don't reproduce very low frequencies very well and if you look at a 300 Hz square wave that was EQ'd with a high pass filter at 30 Hz you should see why. The lower the square wave frequency, the worse the result will get (a 30 Hz square wave will not look like a square wave anymore after filtering). And that's with perfect filters, i.e. without distortion like a driver would produce.
 
Secondly, a 300 Hz square wave has 36 odd harmonics if limited by a sample rate of 44.1 kHz. Even if most power is in the first harmonics, a simple high-shelf filter that cuts the treble a bit (say 5 to 10 dB) changes the shape of the square wave visibly.
 
Another problem is the diffuse field equalization of headphones which is not even being considered here.

 

[bigshot]

The best kind of signal to judge audio fidelity by is *music*.

 

[ultrabike]

Hello gentlemen, nice to be part of HF!

 
Quote:

What about distortion? The reason speakers have so much of a problem reproducing square waves is because the push and pull is so instantaneous, the flapping paper cones can't react fast enough. This causes distortion, making the square wave not very square. The problem isn't the frequency of the square wave, it's the unnatural shape of it.

Well, if you band-limit the square wave to the first 5 tones the pseudo-square wave shouldn't be discontinuous and well within the realm of what the paper cone can do (It'll be curvy)
 
Quote:

But it is not. Headphones and speakers don't reproduce very low frequencies very well and if you look at a 300 Hz square wave that was EQ'd with a high pass filter at 30 Hz you should see why. The lower the square wave frequency, the worse the result will get (a 30 Hz square wave will not look like a square wave anymore after filtering). And that's with perfect filters, i.e. without distortion like a driver would produce.

 
I feel you have a good point about low frequency issues with transducers, so I can see the headphone will show it's true colors with a 30 Hz square wave. That said, a 300 Hz waveform with no DC offset does not have frequency components bellow 300 Hz.
 
Quote:

Originally Posted by xnor /img/forum/go_quote.gif
 
Secondly, a 300 Hz square wave has 36 odd harmonics if limited by a sample rate of 44.1 kHz. Even if most power is in the first harmonics, a simple high-shelf filter that cuts the treble a bit (say 5 to 10 dB) changes the shape of the square wave visibly.
 
Another problem is the diffuse field equalization of headphones which is not even being considered here.

 
If you assume 40 kHz there are 33, if you assume 44.1 kHz it is 36 I guess. And yes, even keeping 31 harmonics will yield a wavy looking square wave. But If you have a reference of how a filtered square wave with 5 elements should look, you could compare it to what you are getting out of the headphone. Some will do better than others, and it should sound that way...
 
The diffuse field equalization and HRTF is yet another issue there for sure (I feel it can be factored in the measurements though through compensation)

 

[jcx]

are people here really this clueless about loudspeaker/room early reflections, low frequency modes, reverberation contribution, decay time?
lots of Linkwitz-Riley 2nd/4th order XO out there with inverting signal polarity 180 or 360 degree phase shift through the XO region
 
you don't get a square wave SPL at the listening position out of typical multidriver loudspeakers in typical domestic rooms with minimal/no attention to speaker placement, dimensions, acoustic treatment
 
and many loudspeaker systems still manage well enough to make discussion of headphone "soundstage", "imaging" ROFLMAO silly exaggeration

 

[ultrabike]

Quote:

The best kind of signal to judge audio fidelity by is *music*.

 
Well, some sound effects can end up being pseudo-band-limited-square-wavy signals that can be present in music... Jimmy Hendrix did some crazy stuff there... not exactly square (tubes), but somewhat close...

 

[bigshot]

It really doesn't matter if synthetic sound becomes distorted, as long as it sounds interesting. But when it comes to accuracy of reproduction, what I'm interested in is symphony orchestras, saxophones, drums,violins, pianos, etc... Those are hard to get to sound right, and you can instantly tell when they're off.

 

[bigshot]

Well, if you band-limit the square wave to the first 5 tones the pseudo-square wave shouldn't be discontinuous and well within the realm of what the paper cone can do (It'll be curvy)

Square waves aren't curvy are they? Perhaps I'm missing something...

 

[ultrabike]

Quote:

are people here really this clueless about loudspeaker/room early reflections, low frequency modes, reverberation contribution, decay time?
lots of Linkwitz-Riley 2nd/4th order XO out there with inverting signal polarity 180 or 360 degree phase shift through the XO region
 
you don't get a square wave SPL at the listening position out of typical multidriver loudspeakers in typical domestic rooms with minimal/no attention to speaker placement, dimensions, acoustic treatment
 
and many loudspeaker systems still manage well enough to make discussion of headphone "soundstage", "imaging" ROFLMAO silly exaggeration

 
I guess there are things headphones do better than speakers, and there are things speakers do better than headphones...
 
Quote:

Square waves aren't curvy are they? Perhaps I'm missing something...

 
A perfect square wave is not curvy and it has infinite bandwidth and odd harmonics. A square wave generator, and an audio amplifier are not infinite bandwidth and will only manage to approximate a square wave, which in the end will look curvy depending on the number of harmonics they can reproduce:
 
http://cnx.org/content/m0041/latest/
 
A headphone is not infinite bandwidth either (nor does it need to be.) It should be accurate within our hearing range... Meaning an ideal SPL measurement for a 300 Hz square wave (after HRTF and other considerations) should measure as a 36 term approximation to a square wave if using 44.1 kHz sampling (per xnor observations)

 

[bigshot]

Sorry to be dense, but to me, it's a matter of relativity. A square wave generator's output is going to look a LOT more like a square wave than it will after a speaker gets done with it, no? The point is, how is accuracy reproducing square waves any prediction of how well a speaker produces music? A speaker can scrunch up a square wave like a beer drinking truck driver does with empty aluminum cans, but that doesn't mean it won't reproduce a Mozart piano concerto with startling presence and clarity.

What kind of instrument requires 36 levels of harmonics? Generally, when equalizing speakers, three or four levels is plenty. Edit: Deleted my sorry attempt at math....

Forgive me if I keep pulling things back down to the real world. It's the way I am...

 

[ultrabike]

Quote:

Sorry to be dense, but to me, it's a matter of relativity. A square wave generator's output is going to look a LOT more like a square wave than it will after a speaker gets done with it, no? The point is, how is accuracy reproducing square waves any prediction of how well a speaker produces music? A speaker can scrunch up a square wave like a beer drinking truck driver does with empty aluminum cans, but that doesn't mean it won't reproduce a Mozart piano concerto with startling presence and clarity.
Forgive me if I keep pulling things back down to the real world. It's the way I am...

 
I'm fairly dense myself

, I'm not who to criticize... I all depends on the fundamental frequency of the square wave. You can manage more harmonic terms of a 300 Hz square wave vs a 10 kHz square wave in a system that can't reproduce anything past 20 kHz... I guess a 10 kHz square wave might sound like a 10 kHz sinusoidal to our ears anyway.
 
I have to say a square wave IMHO is not a very good test compared to a full impulse response measurement. However, if a headphone has severe issues replicating an approximate 300 Hz band-limited pseudo-square wave... well... it is not something to brag about...
 
...
 
BTW, harmonics are really just tones. You should be able to hear at least some of those tones and more. Have you heard white or pink noise? That is a bit more than 36 tones added together there...

 

[bigshot]

That helps a bit... But I'm really not interested in bragging about my headphones. I want to listen to music with them. I don't see how square wave accuracy translates into better reproduction of music. And even if my system could go beyond 20kHz, I don't see how harmonics above that would improve the way my stereo sounds to me. This all sounds like armchair theorizing, and what I am interested in is practical application.

On a similar subect... I see people referring to "speed" of headphones, and then they point to numbers representing slivers of time so small, I can't imagine people being able to perceive them. It makes no sense to me.

 

[xnor]

Quote:

 
That said, a 300 Hz waveform with no DC offset does not have frequency components bellow 300 Hz.

 
I know but the filter still has an effect on higher frequencies and therefore changes the square wave considerably.
 
 
Quote:

Quote:
 
I have to say a square wave IMHO is not a very good test compared to a full impulse response measurement. However, if a headphone has severe issues replicating an approximate 300 Hz band-limited pseudo-square wave... well... it is not something to brag about...

But how do those issues look like? With what I explained above, you'll get a wiggly, non-square wave looking result just from some "perfect" filters (high pass, diffuse field EQ, slight bass boost and treble cut) applied to a square wave.
 
Without analyzing the square wave by transforming it into magnitude, phase response etc. you can only guess what the wiggles are telling, which everyone seems to interpret differently (wrong). Square waves simply are not a very good choice/test signal. Guess that's why most speaker measurement apps don't use/recommend it.

 

[ultrabike]

Quote:

That helps a bit... But I'm really not interested in bragging about my headphones. I want to listen to music with them. I don't see how square wave accuracy translates into better reproduction of music. And even if my system could go beyond 20kHz, I don't see how harmonics above that would improve the way my stereo sounds to me. This all sounds like armchair theorizing, and what I am interested in is practical application.

If you see my profile you will find that the headphones I currently own are not reference levels of neutrality by any means... I would rather have the Stax Lambda 404's, 007's, Joe Grados, HE-500 or even a lower level HD600... which while not perfect, definitively better than what I have currently IMHO. I might get there with time

. I'm also an equalization believer and would love to see a portable stand alone good equalizer for headphones with more functionality than just 10 bands, simple cross-feed, and bass-boost...
 
I do see your point. Music is the end game... but it should be clear I do put some value to measurements there...

 

[ultrabike]

Quote:

 
I know but the filter still has an effect on higher frequencies and therefore changes the square wave considerably.
 
 
But how do those issues look like? With what I explained above, you'll get a wiggly, non-square wave looking result just from some "perfect" filters (high pass, diffuse field EQ, slight bass boost and treble cut) applied to a square wave.
 
Without analyzing the square wave by transforming it into magnitude, phase response etc. you can only guess what the wiggles are telling, which everyone seems to interpret differently (wrong). Square waves simply are not a very good choice/test signal. Guess that's why most speaker measurement apps don't use/recommend it.

 
If you use a 5th order Butterworth maximally flat with corner at 200 Hz things might look better in the digital domain:
 
MATLAB:
[B,A]=butter(5,200/44.1e3,'high');
[H,F]=freqz(B,A,1024,44.1e3);
semilogx(F,20*log10(abs(H)));
grid on;
axis tight;
xlabel('Hz');
ylabel('Magnitude');

 
And I agree, things get complicated if compensating for ear and head there. There needs to be a baseline somewhere, One can do relative measurements then. I'm not saying this is simple at all. Just possible to some degree.