[marcan]

   
Play original musical signal through 2 parametric filters that reverse each other: (1) +3 dB @ 4.5 kHz, Q=0.67, then (2) -3 dB @ 4.5 kHz, Q=0.67. If both EQs were perfect, the final waveform would be identical to the original. 

It won't be check it. You will necessarily add noise in the time domain. If you use a linear phase you will have pre ringing which is particularly unnatural (in the real world you don't hear a sound before it starts). If you use a minimal phase you will have post ringing which is more natural (in the real world you always hear a tail after a sound) but the phase is compromised. Nowadays there is no way around it.
 

I'm willing to try this experiment. I can create 2 WAV files in exactly this way and post them here. Anyone can load them into a DBT app and see if they can tell them apart without guessing (confidence > 95%).
 
To be fair to your point, if we used a more extreme filter - bigger amplitude or steeper slope - then at some point the distortions you mention become audible. But not with slopes at 3 dB / octave.
 

Sure I'm ready to abx

 

[SearchOfSub]

Are you sure you aren't listening to tracks that have noise in the recording?

The Schiit stack should not have a noise floor, but the LCD-2 is so detailed that it can render all the imperfections on a track.

Check out the best recordings you have in your library and see if the noise exists on all of them. It shouldn't.

The Schiit products I have tried both had lot of noise. I label Schiit products as gobs of power & noise nowadays.

 

[droopy1592]

The Schiit products I have tried both had lot of noise. I label Schiit products as gobs of power & noise nowadays.

 
I have a Schiit stack, Modi MB and magni 2 uber and even with ultra resolving phones there is no extra noise. I only hear noise when it's present in the track. 
 
Macbook pro- (both optical and usb compared) modi MB- magni 2 uber no issues no noise
 
Everything from PM3/RS2/RS1/HD800 and S, LCD2.2 and PS500E and HE560
 
no additional noise

 

[franzdom]

The Schiit products I have tried both had lot of noise. I label Schiit products as gobs of power & noise nowadays.

 
Was there a DAC involved or just amps?

 

[ScareDe2]

Are you sure you aren't listening to tracks that have noise in the recording?

The Schiit stack should not have a noise floor, but the LCD-2 is so detailed that it can render all the imperfections on a track.

Check out the best recordings you have in your library and see if the noise exists on all of them. It shouldn't.

Hi
I have try on another computer at home with same result. I tried on my portable same result. I have tried different rock/metal songs same result. I hear it with flac, cd and mp3. The noise is there only when music is playing. I hear it with both LCD2 and DT770.

 

[Malfunkt]

Using EQ within the range of 5db is fine. If you use an EQ that has gain compensation it isn't going to matter if you boost or cut, there should be no distortion. 
 
Now, they will alter the sound and essentially create resonance, messing with formants / partials - noticeable if you are creating sharp peaks at high gain.
 
It should be said, because of the design of certain headphones enclosures and drivers, no amount of EQ may rectify an issue. In some cases, such as the pair of Denon AH-D2000, I can go ahead and EQ it so the bass is very prominent, and crank to loud levels, as it can handle this. Still doesn't sound ideal. Within a modest range is fine, but typically I prefer to listen to all my headphones as they are and let my brain do the adjusting. If something is a bit sharp, I'll just turn the headphone down. Initially after getting the D2000 I thought it was too sharp, but now I have adjusted to it and feel everything else I have is not detailed enough.
 
I guess experiment and use your ears, just make sure you are not boosting a frequency band and causing digital distortion - make sure the gain is compensated or only cut frequencies.
 
http://www.soundonsound.com/techniques/whats-frequency

 

[tp147]

  Yes it will add at least ringing, quantisation noise or phase shift. If you don't hear it it's another story. 

The phase shift will be the wirst thing.
 
I never understood people saying keep the phase in mind while designing X-Overs.
Since I'm used to hear music with full range speakers I understand what they are talking about.
 
But the head phone stays a full range, so maybe it wouldn't affect too much.
 
cheers
 
Thorsten

 

[droopy1592]

 
Hi
I have try on another computer at home with same result. I tried on my portable same result. I have tried different rock/metal songs same result. I hear it with flac, cd and mp3. The noise is there only when music is playing. I hear it with both LCD2 and DT770.

Maybe you have a bad unit. I have been in quiet sections of songs before wondering if there is a disconnect somewhere because the stack is so quiet

 

[MRC001]

  Sure I'm ready to abx

OK here are the files:
Clip A
http://mclements.net/audio/eq-test-A.flac
Clip B
http://mclements.net/audio/eq-test-B.flac
 
One of them has 2 opposite parametric filters applied. The other has no filter or processing at all. Load them into your favorite A/B/X software for a DBT and see if you can reliably tell which one. One of these tracks actually has TWICE the distortion added by a single parametric filter, so it should be easy, right?
 
It's a high quality full bandwidth recording with big dynamics and fast transients, so that should make it even easier! All those fast transients should make even the slightest time smearing, pre or post echo obvious!
 
<evil grin> I think not, I sure can't tell any difference. But I'm willing to be surprised by someone with "golden ears".

 

[marcan]

  OK here are the files:
Clip A
http://mclements.net/audio/eq-test-A.flac
Clip B
http://mclements.net/audio/eq-test-B.flac
 
One of them has 2 opposite parametric filters applied. The other has no filter or processing at all. Load them into your favorite A/B/X software for a DBT and see if you can reliably tell which one. One of these tracks actually has TWICE the distortion added by a single parametric filter, so it should be easy, right?
 
It's a high quality full bandwidth recording with big dynamics and fast transients, so that should make it even easier! All those fast transients should make even the slightest time smearing, pre or post echo obvious!
 
<evil grin> I think not, I sure can't tell any difference. But I'm willing to be surprised by someone with "golden ears".

First test and I'm a bit tired it's 6:30am before sleep
Easy. I'm sure I can do 8/8 with a bit of sleep.
 
foo_abx 2.0.2 report
foobar2000 v1.3.10
2016-09-17 06:31:42
File A: eq-test-A.flac
SHA1: 3e6b72bc48b5a9e16360cd3f45c03fa12b3c1d21
File B: eq-test-B.flac
SHA1: 003ebe932c126bd329d977422b4ea150cadf1696
Output:
ASIO : ASIO Fireface USB
Crossfading: NO
06:31:42 : Test started.
06:31:53 : 01/01
06:32:29 : 01/02
06:32:45 : 02/03
06:32:56 : 03/04
06:33:07 : 04/05
06:33:22 : 04/06
06:33:30 : 05/07
06:33:56 : 06/08
06:33:56 : Test finished.
 ---------- 
Total: 6/8
Probability that you were guessing: 14.5%
 -- signature -- 
2b10e5f4bd93ad0cb42c6d59f56f96493ee7e08d

 

[MRC001]

Thanks for playing. Two points:
6 of 8 is only 85.5% confidence. A valid result is 95% or higher.
Telling them apart is nice but not sufficient - which has the filters?

 

[marcan]

Thanks for playing. Two points:
6 of 8 is only 85.5% confidence. A valid result is 95% or higher.
Telling them apart is nice but not sufficient - which has the filters?

 
No problem I will do 8/8 after sleep.
Did you do the test?

 

[MRC001]

Sure I did. I can tell them apart with 95% confidence only when the ABX switch delay is zero, because they're not perfectly time aligned. So the difference I'm hearing isn't ripple or phase or any kind of distortion, it's simply time shift. How do I know? When I set the ABX switch delay to 100 msecs I cannot tell them apart with 95% confidence.

Good luck! The big question: which has the filters applied?

 

[ScareDe2]

  Maybe you have a bad unit. I have been in quiet sections of songs before wondering if there is a disconnect somewhere because the stack is so quiet

Hi
it seems that those 90's rock/metal bands were using compression maybe that is what sound weird at about 5kHz. I can't be sure yet I am learning. thanks for the help.

 

[Lohb]

 
Hi
it seems that those 90's rock/metal bands were using compression maybe that is what sound weird at about 5kHz. I can't be sure yet I am learning. thanks for the help.

Get some excellently mastered stuff to try. There is a website posted on here or another Audeze thread that rates mastering quality of albums. I'd put it down to poor mastering, it is not something that comes up a lot.