• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Does Phase Distortion/Shift Matter in Audio? (no*)

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
But I am pretty sure that there is no velocity field, and that it is only a pressure field… with the speed of propagation being a constant at any temperature.
Velocity field means particle velocity field here, as @René - Acculution.com has already pointed out (IIRC). The complete sound field at a point in space is described by a 3D velocity field vector and a scalar pressure component
And it can be directly measured with a hot wire microphone, besides the indirect method of measuring pressure gradients. The "wind" from the particles in motion removes heat from a hot wire by convection cooling, which in turn can be measured via the resistance change of the wire. Requires very small/thin wires to get to high enough frequecies, though. The technology is more than 100 years old.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,013
Location
Berlin, Germany
Regardless how "good" are the crossover filters, we still have to follow basic laws of acoustics and see the directivity depending on distance of sources, frequency and number of sources. Plots and formula below are for point sources, number of them = n, distance = d and Lambda is the wavelength. View attachment 237392No crossover.

From Josef Merhaut, Theory of Electroacoustics
We also have XDir, a handy little simulator based on above formulas, again for point sources with uniform directivity, flush on an infinite baffle.
 

NTK

Major Contributor
Forum Donor
Joined
Aug 11, 2019
Messages
2,660
Likes
5,820
Location
US East
If the design of crossover is not good enough due to cost limitation,
can rePhase improve the sound quality?

I think the answer is yes if we can measure the phase introduced in the crossover directly.
But what most people here measured is the phase at the MLP.
I have never used rePhase. Based on my limited understanding, and assuming you are referring to using rePhase to manipulate the signal which is then fed into speakers with passive crossovers, I don't think so.

At the crossover frequency, the highpass and lowpass filters in the passive crossover will alter the phase differently and independently. Since the input to the crossover filters is the same, there will be phase differences in the outputs of the crossover filters. If the speaker crossover is not properly designed, which causes directivity errors, manipulating the signal at the input won't be able to correct it.

IME, by manipulating the input signal, you should be able to get a "correct" response at one single axis /direction (but will need to watch out for cancelation dips as simple inverse algorithms will simply apply huge amounts of gain to counter the dips).
 

gnarly

Addicted to Fun and Learning
Joined
Jun 15, 2021
Messages
992
Likes
1,390
I have never used rePhase. Based on my limited understanding, and assuming you are referring to using rePhase to manipulate the signal which is then fed into speakers with passive crossovers, I don't think so.

At the crossover frequency, the highpass and lowpass filters in the passive crossover will alter the phase differently and independently. Since the input to the crossover filters is the same, there will be phase differences in the outputs of the crossover filters. If the speaker crossover is not properly designed, which causes directivity errors, manipulating the signal at the input won't be able to correct it.

IME, by manipulating the input signal, you should be able to get a "correct" response at one single axis /direction (but will need to watch out for cancelation dips as simple inverse algorithms will simply apply huge amounts of gain to counter the dips).
I think you are totally correct.. I have tried rePhase and other phase linearization software, on speakers with either passive or active crossovers already in place, makes no difference..
As you say, doing such can make a "correct" response, but only to a single point in space. (it can't even maintain "correct" response to a single, axis ime....distance matters too.)
 

gnarly

Addicted to Fun and Learning
Joined
Jun 15, 2021
Messages
992
Likes
1,390
We also have XDir, a handy little simulator based on above formulas, again for point sources with uniform directivity, flush on an infinite baffle.
Yes, such a nice easy tool. Many folks like VituixCad for this too. https://kimmosaunisto.net/

Although I'm an advocate ditching the idea of a flat baffle !! ...no need to spread things out so ..:D

Here's two 10" drivers, four 4"s, and a coax CD all tightened up together. All xover summation ranges are well within 1/4 WL.
It's about 300mm between the two 10" drivers' ports, to give a sense of scale.


syn10 ports close.jpg
 

valerianf

Addicted to Fun and Learning
Joined
Dec 15, 2019
Messages
691
Likes
442
Location
Los Angeles
Amir said: "Many crimes are committed in the name of correcting phase/timing in audio such as low order crossovers, stepped drivers"
Well, today trend is to use a wave guide for the tweeter, mechanically pushing it backward.
Is it not the same phase alignment than a stepped cabinet for the tweeter (80's tech)?
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
IME, by manipulating the input signal, you should be able to get a "correct" response at one single axis /direction (but will need to watch out for cancelation dips as simple inverse algorithms will simply apply huge amounts of gain to counter the dips).

In other words, one should consider the overall design and purpose of a speaker and base the EQ beyond a simplistic single point measurement technique.

1665956710680.png




Naive individuals probably think this speaker as atrocious...
 

Holmz

Major Contributor
Joined
Oct 3, 2021
Messages
2,018
Likes
1,241
Location
Australia
Velocity field means particle velocity field here, as @René - Acculution.com has already pointed out (IIRC). The complete sound field at a point in space is described by a 3D velocity field vector and a scalar pressure component
And it can be directly measured with a hot wire microphone, besides the indirect method of measuring pressure gradients. The "wind" from the particles in motion removes heat from a hot wire by convection cooling, which in turn can be measured via the resistance change of the wire. Requires very small/thin wires to get to high enough frequecies, though. The technology is more than 100 years old.

Yeah - but the mass airflow sensor (MAF) and other description of bulk air movement seemed like a departure from phase.
It makes sense for a subwoofer, but the velocity direction of a point source is trivial compared to higher frequency.


Amir said: "Many crimes are committed in the name of correcting phase/timing in audio such as low order crossovers, stepped drivers"

I am not sure that is the biggest crime of the century, but everyone comes at the problem with biases and varying approaches.
What metric(s) would one use to judge whether it is leaning towards good or criminal?
(A step function response? Or impulse response? Or something else?)

There are pages of posts arguing that phase doesn’t matter, but it does not make a signal have higher fidelity to launch it out of a speaker with the phase intentionally jacked up.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
I'm an admitted "phase-oholic", who strives for flat phase (constant delay) in all my DIY speaker builds. (Flat magnitude is taken as a given.)
I truly think i hear superior results from doing so, but cannot say the results are solely/mainly due to achieving flat phase.

A big part of the excellent results may simply be how much easier it is to tune a speaker with complementary linear phase xovers, than with traditional IIR xovers.
Are you equating "flat" phase with "linear phase" (and thus constant group delay)? A question, not a challenge, as the two were treated differently in my past. I have gotten tripped up before by some speaker sites that equate "flat phase" with "0 degrees" just to further confound my little pea brain -- like this one, first in a search: https://www.baffless.com/flat-frequency-and-phase .

Curious, thanks - Don
 

gnarly

Addicted to Fun and Learning
Joined
Jun 15, 2021
Messages
992
Likes
1,390
Are you equating "flat" phase with "linear phase" (and thus constant group delay)? A question, not a challenge, as the two were treated differently in my past. I have gotten tripped up before by some speaker sites that equate "flat phase" with "0 degrees" just to further confound my little pea brain -- like this one, first in a search: https://www.baffless.com/flat-frequency-and-phase .

Curious, thanks - Don
Hi Don, yes, by "flat" phase i mean a level phase trace at zero degrees across the spectrum, when all constant delay has been removed.

I get tripped up by the terms sometimes too.
Whether it's called flat phase of linear phase, in the end i think they mean the same thing...with flat phase just being the special case of constant delay being removed.

My personal opinion is that phase traces make no sense or provide useful info, until constant delay is removed anyway.....which thankfully seems to ne more the norm in displaying phase traces.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
Hi Don, yes, by "flat" phase i mean a level phase trace at zero degrees across the spectrum, when all constant delay has been removed.

I get tripped up by the terms sometimes too.
Whether it's called flat phase of linear phase, in the end i think they mean the same thing...with flat phase just being the special case of constant delay being removed.

My personal opinion is that phase traces make no sense or provide useful info, until constant delay is removed anyway.....which thankfully seems to ne more the norm in displaying phase traces.
Got it, thanks! Yes, watching all the flipping in most phase plots doesn't usually help much...
 

anphex

Addicted to Fun and Learning
Forum Donor
Joined
May 14, 2021
Messages
662
Likes
870
Location
Berlin, Germany
Another question since I had a not unheated argument in a German board: is Phase Shift over the frequency range considered distortion or is it a whole different thing? He said it was just the same as distortion, I said it's a mathematical change of the signal, not added harmonics.

:|
 

René - Acculution.com

Senior Member
Technical Expert
Joined
May 1, 2021
Messages
416
Likes
1,221
Another question since I had a not unheated argument in a German board: is Phase Shift over the frequency range considered distortion or is it a whole different thing? He said it was just the same as distortion, I said it's a mathematical change of the signal, not added harmonics.

:|
It is linear distortion, yes. Often when distortion is mentioned, it is non-linear distortion that is being discussed. I go through it all in my video
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,424
Likes
4,030
Location
Pacific Northwest
Another question since I had a not unheated argument in a German board: is Phase Shift over the frequency range considered distortion or is it a whole different thing? He said it was just the same as distortion, I said it's a mathematical change of the signal, not added harmonics.
It is linear distortion, yes. Often when distortion is mentioned, it is non-linear distortion that is being discussed. ...
Seems right to me. As I see it, any change in the signal is either distortion or noise. If the change is related to or depends on the signal (added harmonics, intermodulation, non-flat frequency response) then it's distortion. Otherwise it's noise.

By this definition, a volume knob creates distortion: a change in signal amplitude. But that is linear and affects all frequencies equally so we disregard it. However, "linear" doesn't necessarily mean "transparent" so I wouldn't disregard all forms of linear distortion. Like non-flat frequency response could be linear but since it affects some frequencies more than others, it's "damaging" the signal in a way that a volume knob does not.
 

René - Acculution.com

Senior Member
Technical Expert
Joined
May 1, 2021
Messages
416
Likes
1,221
Seems right to me. As I see it, any change in the signal is either distortion or noise. If the change is related to or depends on the signal (added harmonics, intermodulation, non-flat frequency response) then it's distortion. Otherwise it's noise.

By this definition, a volume knob creates distortion: a change in signal amplitude. But that is linear and affects all frequencies equally so we disregard it. However, "linear" doesn't necessarily mean "transparent" so I wouldn't disregard all forms of linear distortion. Like non-flat frequency response could be linear but since it affects some frequencies more than others, it's "damaging" the signal in a way that a volume knob does not.
Volume is a constant level change and that does not count as distortion from a signal processing stand point. The same goes for a pure time delay. It is all in the video.
 

j_j

Major Contributor
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
2,267
Likes
4,759
Location
My kitchen or my listening room.
Volume is a constant level change and that does not count as distortion from a signal processing stand point. The same goes for a pure time delay. It is all in the video.

It's been difficult to get the difference between linear functions and nonlinear processes across a few times. Even phase shift and changes in frequency response are linear processes, which is why you can back them out.

As to moving the mike, of course it will change. But how much of it is pure delay?
 

mitchco

Addicted to Fun and Learning
Audio Company
Joined
May 24, 2016
Messages
640
Likes
2,397
What happens if you move the microphone, say, five feet in any direction?
Hey @Flaesh , actually, still really good. Here are 6 measurements of left and right channels. One set at the LP, one set with the mic 3ft to the left of center and another set 3ft right of center. Basically covering my 3 seat couch area:

JBL 4722 w Rythmik dual 18 subs 6 measures.jpg


Don't mind the steep slope, I was playing around listening to different targets.

JBL 4722 w Dual Rythmik F18 subs across 6ft at LP step.jpg


Just a bit of preringing, not audible in my listening tests. Audiolense DRC provides excellent frequency and time domain control, over a wide listening area, even if using a single point analysis measurement. While Audiolense can do multiseat measurements, I find the single point analysis measurement is less of a compromise and sounds better to my ears. Certainly better than the uncorrected time domain mess.
 

Cars-N-Cans

Addicted to Fun and Learning
Joined
May 19, 2022
Messages
819
Likes
1,009
Location
Dirty Jerzey
Hey @Flaesh , actually, still really good. Here are 6 measurements of left and right channels. One set at the LP, one set with the mic 3ft to the left of center and another set 3ft right of center. Basically covering my 3 seat couch area:

View attachment 249958

Don't mind the steep slope, I was playing around listening to different targets.

View attachment 249960

Just a bit of preringing, not audible in my listening tests. Audiolense DRC provides excellent frequency and time domain control, over a wide listening area, even if using a single point analysis measurement. While Audiolense can do multiseat measurements, I find the single point analysis measurement is less of a compromise and sounds better to my ears. Certainly better than the uncorrected time domain mess.
Out of curiosity, what kind of corrections are required on the DSP processing end to get the orderly frequency and impulse response? Seems like it could take quite a bit to hammer out the messiness and all of the room modes in both the frequency and time domain, but I guess if you have capable speakers and plenty of power on tap its an academic discussion.
 
Top Bottom