How do you tell when a headphone is producing a neutral sound?
Dec 6, 2012 at 11:33 AM Post #16 of 31
"Correcting the non-linearities"?

That kind of language makes me nervous. Equalization is correcting linear features like frequency and phase response, via a linear operation (multiplication in frequency domain, or convolution in time domain; or what is the actual implementation in practical audio equalizers? in hardware you can do an RLC network with a certain transfer function, which we're hoping has few nonlinearities)

The nonlinearities are exhibited in features like harmonic distortion and resonances, which are not being directly compensated for.

Or am I missing something here?
 
Dec 6, 2012 at 11:58 AM Post #17 of 31
Quote:
The only way to achieve linear-perceived response on your headphones is to use non colored, non flattering convertors (DACs), amplifiers and a (preferably linear phase) hi quality algorithm equalizer.

I agree with most of what you wrote, but linear-phase means pre-ringing. For the equalization of dynamic drivers you want minimum phase, i.e. there's no energy before an impulse and most of the energy is concentrated near the start of an impulse. That way you can not only correct the magnitude but also the phase of the frequency response.
 
As for "hi quality". Imo an EQ either works properly or it doesn't. Even free ones usually operate with 64-bit floating point and use textbook formulas that work perfectly. Things get hairy when the developers think they have to add hidden filters or oversampling or other special "features".
 
Dec 6, 2012 at 12:24 PM Post #20 of 31
The important thing to understand about neutrality in sound is that it is not something that is debatable or subjective. It simply means linear (flat) and that is all there is to it. If a headphone measures flat, it is neutral. There are not many headphones that achieve this ideal flatness, but some come pretty close.
 
Dec 6, 2012 at 12:59 PM Post #21 of 31
Dec 6, 2012 at 1:46 PM Post #22 of 31
Quote:
I found out the term "neutral" only in non professional audio communities. This is because the usual consumer and even hi end (read expensive) products are very "colored".

 
Most consumer CD players, receivers and amps are totally flat. It isn't hard to manufacture electronics to spec. The difficulty comes with headphones and speakers and how they react with the rooms they're in.
 
Dec 6, 2012 at 2:18 PM Post #23 of 31
Quote:
One good way of doing it is to listen to voice, or orchestra if you're familiar with how they sound live. As others stated, no frequency is accentuated or recessed. I don't think this is a very accurate way of telling though, as brains will "learn" a specific headphones sound as accurate, then when you switch over to something that actually is accurate, it will sound very "off" until your mind "re-learns" how things should sound. Another way of telling of course, is by looking at frequency response graphs (provided you can find them for your particular headphones). Keep in mind however you will not see a straight line from 20hz-20khz because headphones have to compensate for your ears natural equalization of frequencies heard at a distance. Here are a few example of headphones that are reasonably accurate.
 

None of these headphones are 100% completely accurate, but they do come pretty damned close. None of them have accurate bass either. Theoretically the line should remain flat at 0dB from 10hz on to around 1000hz

This is a great explanation IMO
 
Dec 6, 2012 at 2:38 PM Post #24 of 31
The trick to EQing using recordings of natural sounding acoustic instruments is to make very small corrections and refine the response curve over weeks or months using a wide variety of accurate recordings. I try not to change more than 1.5 dB at a time, so I don't parallel park too much. It helps me keep focused on where I am and where I want to be.

Classical music is ideal for tuning your system because it includes the whole spectrum of sound, not just midrange like vocals or instrumental music. Piano music can be helpful to identify bumps or dips as the performer does runs across the keyboard.
 
Dec 6, 2012 at 2:49 PM Post #25 of 31
Nevermind I understood my mistake
 
Dec 6, 2012 at 2:49 PM Post #26 of 31
Quote:
"Correcting the non-linearities"?
That kind of language makes me nervous. Equalization is correcting linear features like frequency and phase response, via a linear operation (multiplication in frequency domain, or convolution in time domain; or what is the actual implementation in practical audio equalizers? in hardware you can do an RLC network with a certain transfer function, which we're hoping has few nonlinearities)
The nonlinearities are exhibited in features like harmonic distortion and resonances, which are not being directly compensated for.
Or am I missing something here?

 
Sorry about that. As you can see English is obviously not my primary language. Should I say peaks and valleys? I was obviously referring to frequency response.
Quote:
I agree with most of what you wrote, but linear-phase means pre-ringing. For the equalization of dynamic drivers you want minimum phase, i.e. there's no energy before an impulse and most of the energy is concentrated near the start of an impulse. That way you can not only correct the magnitude but also the phase of the frequency response.
 
As for "hi quality". Imo an EQ either works properly or it doesn't. Even free ones usually operate with 64-bit floating point and use textbook formulas that work perfectly. Things get hairy when the developers think they have to add hidden filters or oversampling or other special "features".

 
It really depends on the algorithm implemented in that linear phase eq. Thats why some sound super transparent (and are equally expensive, like the Algoritmix Red) while others don't.
 
Hi quality can mean oversampling to avoid aliasing, but also "analog" phase algorithms (like in EQuality) or analog like curves (like those in the Eiosis Eq). To me the "extreme" setting in the iOS app named "Equalizer Pro" sounds superior to the sound of another excellent app named "Equalizer" which also has a full parametric eq. But I prefer to fiddle with the interface of Eq Pro for that.
 
Actualy for listening to the Tesla T1 I admit I use a minimum phase eq which is included with the Metric Halo ULN-2 I use at home, but the mixer on that has 80bit resolution, and the eq sounds great. You gave me a reason to test minimum/linear phase eq vs dynamic/ba based phones 
L3000.gif

 
Dec 6, 2012 at 3:02 PM Post #27 of 31
For a given magnitude response there is only one proper way a minimum- or linear-phase EQ has to work. I see no reason for an EQ to be expensive.
Linear-phase only is transparent if you do small/smooth adjustments, else you'll get audible pre-ringing.
 
Dec 9, 2012 at 9:55 PM Post #30 of 31
Quote:
for me its when not one type ofgenre souunds good and you can tell this is how the song was meant to be heard by the artist.

 
Problem is how to really tell if that's how the artist meant it to be heard. If for example one of the artists that dominate my playlists uses this in the studio and this as reference for home speakers, I can't confirm what "they meant to be heard" until I try these (and these aren't just sitting on display in just any store), short of just buying precisely those (which I cant afford even if I sell my car and one of my kidneys). Which of course is still a substitute for the composer to visit my home (or kidnap him) and tell me what's wrong with it based on how he mastered the CDs.
 

Users who are viewing this thread

Back
Top