Head-Fi.org › Forums › Equipment Forums › Sound Science › Does frequency response (or CSD) entirely determine sound quality?
New Posts  All Forums:Forum Nav:

Does frequency response (or CSD) entirely determine sound quality?

post #1 of 26
Thread Starter 

Let me make a bold statement here:

If we have two different headphones with the same CSD plots, it will be not possible to tell a difference between them in blind testing.

 

Can someone prove me wrong? Are there other scientifically proven properties beside waterfall plots which can influence how sound equipment will sound? Because if there are not, then with Occam's razor in mind, we can assume that my statement is true.

 

If that's true, then if someone invents a relatively flat sound headphones with very quick decay across whole FR range, it could be reshaped to any other existing headphones with precise EQ (a longer decay could be artificially added by software).

 

I hope I don't sound like an ignorant, I'd really like to learn more about sound science. The first thing that comes into my mind is the interference between frequencies, for example if there is a constant sine wave at some frequency point it could also change response of others, however I haven't found any article proving this.

post #2 of 26
Quote:
Originally Posted by ieee754 View Post
 

Let me make a bold statement here:

If we have two different headphones with the same CSD plots, it will be not possible to tell a difference between them in blind testing.

 

Can someone prove me wrong? Are there other scientifically proven properties beside waterfall plots which can influence how sound equipment will sound? Because if there are not, then with Occam's razor in mind, we can assume that my statement is true.

 

If that's true, then if someone invents a relatively flat sound headphones with very quick decay across whole FR range, it could be reshaped to any other existing headphones with precise EQ (a longer decay could be artificially added by software).

 

I hope I don't sound like an ignorant, I'd really like to learn more about sound science. The first thing that comes into my mind is the interference between frequencies, for example if there is a constant sine wave at some frequency point it could also change response of others, however I haven't found any article proving this.

 

The two issues with frequency response are basically:

1. it doesn't address the time domain

2. it assumes headphones are linear

 

1 is mostly not that important. Except for bass frequencies, the natural decay time of instruments is much longer than that of the headphone itself.

 

You can see the effects of 2 mostly in harmonic distortion, which also primarily affects bass frequencies.

 

 

Occam's razor is a principle; it doesn't prove anything. Just because we mostly know about something doesn't mean we know everything. Though science is very good at knowing how far we understand.

post #3 of 26

You forgot distortion.

post #4 of 26
Thread Starter 

I know about distortions, however I've read somewhere that THD below 1% is inaudible, and since most modern high-end headphones are far below that level I assume that it doesn't influence the sound.

post #5 of 26
Thread Starter 

Another thing I can think of is how fast headphone reacts to signal change, and probably the square wave response is the best measurement for this. However I think that output of these measurements should be compensated with the Fourier series based on headphone FR graph which represents how square wave will be deformed if it reacts with zero delay (otherwise it is nothing more than compact FR plot).


Edited by ieee754 - 3/22/14 at 6:16pm
post #6 of 26

It's very complicated. I'd like to add that we have to factor the amp's damping factor caused by initial resistance of headphones and output resistance of the amplifier. If damping is not enough, the frequency will change.

post #7 of 26
Quote:
Originally Posted by ieee754 View Post
 

I know about distortions, however I've read somewhere that THD below 1% is inaudible, and since most modern high-end headphones are far below that level I assume that it doesn't influence the sound.

You can assum that a single scalar value for THD sums up everything about distortion. You can assume that phase linearity is in audible. You can assume that the assumptions youve made are correct, then you'd be correct (as long as your assumptions hold!). 

 

Be careful with occums razor. You should use it when formulating testable hypotheses, not for validating a false premise

 

Cheers

post #8 of 26
It isn't a matter of absolutes, it's a matter of scale. Differences in frequency response can make massive differences in sound quality, and different response curves between different brands and models of cans are very common.

Time issues, distortion and all of the other aspects that go into making headphones sound the way they do aren't completely unimportant, but in the grand scheme of things, they are an order of magnitude before response in importance.

Rather than asking, "Do two different headphones with the same response curve sound the same?" your second question is better... "Can headphones be equalized to sound the same?" In many cases, the answer to that is yes. If cans have the latitude to have their response pushed around a little without overdriving, and the headphones you are matching to have a pretty natural response, the answer is definitely yes. Overdriving to match a wildly different response would be a no.

Midrange headphones can be equalized to sound as good as much better cans. The same is even more true for speakers.
Edited by bigshot - 3/23/14 at 3:53am
post #9 of 26
Quote:
Originally Posted by ieee754 View Post

Let me make a bold statement here:
If we have two different headphones with the same CSD plots, it will be not possible to tell a difference between them in blind testing.

Can someone prove me wrong? Are there other scientifically proven properties beside waterfall plots which can influence how sound equipment will sound? Because if there are not, then with Occam's razor in mind, we can assume that my statement is true.

If that's true, then if someone invents a relatively flat sound headphones with very quick decay across whole FR range, it could be reshaped to any other existing headphones with precise EQ (a longer decay could be artificially added by software).

I hope I don't sound like an ignorant, I'd really like to learn more about sound science. The first thing that comes into my mind is the interference between frequencies, for example if there is a constant sine wave at some frequency point it could also change response of others, however I haven't found any article proving this.
I think you are spot on With that assumption if we take THD out of the Equation. The more EQ and DSP you add, the higher the THD values will normally rise. If we say most People prefer CSD plot X, its generally easier to make the raw measuring headphone perform as Close to X as we can, and then apply crossfeed and other DSP effects to enchance the experiance, as opposed to start from a totally uniformely flat measurement and apply rigorous modifications to get it to perform in a way thats not Natural for the original Construction to do.
post #10 of 26
Quote:
Originally Posted by MatsGyver View Post

The more EQ and DSP you add, the higher the THD values will normally rise.

 

That may be true, but if the levels are still below the threshold of audibility, which is almost always the case, it doesn't matter a bit. A little more distortion you can't hear doesn't mean jack diddly when you are getting more balanced response from equalizing and better room acoustics from DSPs. Equalization and DSPs aren't distortions of the original sound, they're corrections to problems that bring you closer to the original sound.

post #11 of 26
Quote:
Originally Posted by MatsGyver View Post


The more EQ and DSP you add, the higher the THD values will normally rise.

A linear DSP or digital EQ will not add harmonic distortion; it's mathematically impossible. At most they might raise the noise floor. An analog device definitely can and will add to THD, though. 


Edited by higbvuyb - 3/23/14 at 10:46pm
post #12 of 26
Raising the frequency band will raise the distortion in the headphone.
post #13 of 26
Thread Starter 
Quote:
Originally Posted by MatsGyver View Post

Raising the frequency band will raise the distortion in the headphone.

If you raise it to the clipping level then that's true. But EQ can be performed by lowering all bands (and leaving highest at 0 level).

 

Quote:
Originally Posted by higbvuyb View Post
 

A linear DSP or digital EQ will not add harmonic distortion; it's mathematically impossible. At most they might raise the noise floor. An analog device definitely can and will add to THD, though. 

That's interesting, I thought it's the opposite - that analog EQ is more precise as it operates on almost-continuous data (flow of electrons), and digital EQ could introduce errors by FPU inaccuracy.

post #14 of 26
Quote:
Originally Posted by ieee754 View Post
 

If you raise it to the clipping level then that's true. But EQ can be performed by lowering all bands (and leaving highest at 0 level).

 

That's interesting, I thought it's the opposite - that analog EQ is more precise as it operates on almost-continuous data (flow of electrons), and digital EQ could introduce errors by FPU inaccuracy.

Floats (singles) use 24 bits to hold the argument and the final 8 for the exponent. meanwhile, audio data fits into 16 or 24 bit integers, with hardware capable of reproducing it at best, about 22 ENOB (effective noumber of bits) . What sort of eigenvalues does a typical eq filter have? Either way, since only the 16 most signifcant bits are going to matter, id be awfully surprised if a good digital eq would lose 8 bits of precision to truncation error.

 

Cheers

post #15 of 26
Quote:
Originally Posted by ieee754 View Post
 

If you raise it to the clipping level then that's true. But EQ can be performed by lowering all bands (and leaving highest at 0 level).

 

This only really works with graphic equalizers (and then some badly designed ones might not have a flat response if all bands are set to the same non-0 dB level). A better approach that is typically used by parametric equalizers is to provide an overall gain control that can be used to attenuate the signal to avoid clipping.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Does frequency response (or CSD) entirely determine sound quality?