• 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*)

BrEpBrEpBrEpBrEp

Active Member
Joined
May 3, 2021
Messages
201
Likes
245
2 videos and 8 pages discussing the audibility of phase later and nobody's posted a single blind ABX test result? I'll get things started:

View attachment 134830

Pretty obvious to me. Much more than any difference in SINAD between two non-broken DACs/amps (i.e. inaudible).
This thread started out talking about phase distortion from DACs, rather than polarity inversion as you've tested here. Though the topic seems to be constantly changing as it goes.
 
Last edited:

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
Which is why I posted earlier about some of the problems in the conversation.
Just using the word “phase” without context or qualification isn’t useful. Statements claiming that phase is or isn’t audible aren’t even wrong. They are vacuous. Without being clear what we mean it is just a waste of time.
Clearly some aspects of phase change in the signal are audible. Others are not.
The video from PS Audio is where we came in. It is a good example of not even wrong. There is so much confusion of terminology, lack of understanding of auditory mechanisms and general fuddle headedness that Amir only scratched the surface of how awful it was.
The trick here is to not go down the same path.
 
Last edited:

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
6,874
Likes
6,672
Location
UK
Which is why I posted earlier about some of the problems in the conversation.
Just using the word “phase” without context or qualification isn’t useful. Statements claiming that phase is or isn’t audible aren’t even wrong. They are vacuous. Without being clear what we mean it is just a waste of time.
Clearly some aspects of phase change in the signal are audible. Others are not.
The video from PS Audio is where we came in. It is a good example of not even wrong. There is so much confusion of terminology, lack of understanding of auditory mechanisms and general fuddle headedness that Amir only scratched the surface of how awful it was.
The trick here is to not go down the same path.
I watched the video a few days ago & I'm trying to remember the context in which Phase was being discussed. Wasn't it just audibility of phase shifts based on frequency roll off in amplifiers? So ideally we'd be limiting phase discussions in this thread to that usage case? Otherwise things like phase differences in speaker/subwoofer setups can matter, and as another example of how phase can matter GaryH a few posts above this was able to detect music samples that had their absolute phase flipped in blind testing......so phase has a lot of different contexts, I'm certainly no expert on phase, I find it the hardest part of audio to really fathom.
 
Last edited:

Francis Vaughan

Addicted to Fun and Learning
Forum Donor
Joined
Dec 6, 2018
Messages
933
Likes
4,697
Location
Adelaide Australia
Wasn't it just audibility of phase shifts based on frequency roll off in amplifiers?
It was. But the video committed the sin of confusing all uses of “phase” together in a morass of half truths and total rubbish. In particular citing a goodly bunch of phase related issues that have nothing to do with the question he was asked. We are treading the same ground. We do need to be more clear.
 

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
6,874
Likes
6,672
Location
UK
It was. But the video committed the sin of confusing all uses of “phase” together in a morass of half truths and total rubbish. In particular citing a goodly bunch of phase related issues that have nothing to do with the question he was asked. We are treading the same ground. We do need to be more clear.
Yes, you mean Paul's video rather than Amir's (to be clear). Perhaps then we should make sure this thread stays on the topic of phase influences from amplifier frequency roll off then. I think it's safe to say in general that "Phase Matters", but it depends on the context and where in your audio chain it's happening....it's not significant at the amplifier stage when related to frequency roll off of the amplifier.
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
2 videos and 8 pages discussing the audibility of phase later and nobody's posted a single blind ABX test result? I'll get things started:

View attachment 134830

Pretty obvious to me. Much more than any difference in SINAD between two non-broken DACs/amps (i.e. inaudible).

Interesting. How would you describe the difference you hear between inverted and non-inverted? Are there any particular features in the recording that make the inversion obvious for you?
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,274
Location
Gold Coast, Queensland, Australia
2 videos and 8 pages discussing the audibility of phase later and nobody's posted a single blind ABX test result? I'll get things started:

Cool site, thanks for posting, there's some real eye-openers in his various tests. :)
 

Compact_D

Member
Joined
Jun 10, 2021
Messages
47
Likes
17
Nice explanation. But isn't there a possibility of wide spectrum signal like cymbals be spatially "smeared" due to different phase for different frequencies?
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
Nice explanation. But isn't there a possibility of wide spectrum signal like cymbals be spatially "smeared" due to different phase for different frequencies?

Depends whether the group delay is caused by the direct signal leaving the speaker or the room. Toole gives a summary of the scientific literature in his book "Sound production".
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
Perhaps then we should make sure this thread stays on the topic of phase influences from amplifier frequency roll off then.
It is best to look at a few examples to get a sense for the phase shift.

Here would be a roll-off 1st order at 35kHz (FR and phase in green)
1623332416123.png


and here a roll-off 8th order at 35kHz (FR and phase in green)
1623332430751.png


The phase shift that would be caused by the amplifier roll-off is 15° at, for example, 10kHz with the first order filter and 80° with the eighth order filter.

However, the phase shift itself does not express a time delay. Only the group delay, as gradient of the phase curve, has a unit of time.
Even the group delay is usually not equivalent to a time delay (exceptions: pure time delay, allpass).
Source: Testing Loudspeakers, D'Appolito (quoted in substance)

In the examples it can be seen that the group delay is nearly 0, about 0.02ms for the eighth order filter, and is constant over the entire audible frequency range 20-20000Hz:
1623334093121.png
Even if one assumes that the increased group delay corresponds to a time delay, this would be constant and thus inaudible.
Or are there studies that come to a different conclusion?
.
 

GaryH

Major Contributor
Joined
May 12, 2021
Messages
1,348
Likes
1,804
This thread started out talking about phase distortion from DACs, rather than polarity inversion as you've tested here. Though the topic seems to be constantly changing as it goes.

Which is why I posted earlier about some of the problems in the conversation.
Just using the word “phase” without context or qualification isn’t useful. Statements claiming that phase is or isn’t audible aren’t even wrong. They are vacuous. Without being clear what we mean it is just a waste of time.
Clearly some aspects of phase change in the signal are audible. Others are not.
The video from PS Audio is where we came in. It is a good example of not even wrong. There is so much confusion of terminology, lack of understanding of auditory mechanisms and general fuddle headedness that Amir only scratched the surface of how awful it was.
The trick here is to not go down the same path.

Polarity and phase are intimately related. You yourself said:
Yes a polarity inversion is a 180º phase shift.

Absolute phase gets may people riled up. There is almost no evidence it matters, unless something is being driven into some interesting non-linearity. At high enough levels even your ears become non-linear enough that absolute phase can change the way they distort. So you can hear a difference. But unless you do something silly, like connect one channel with reverse polarity, it doesn't matter.

As at least one form of absolute phase change is clearly audible (to me at least) with real music as I've shown, then it certainly does matter.

If there is any serious interest I could provide convolution kernels with different phase distortions so everybody can make their own tests, ABX's or whatever, with any music material. Again, preferably on a system that already is linear phase.

Yes please! ABX tests are the only scientific way to difinitively settle audibility debates.

Interesting. How would you describe the difference you hear between inverted and non-inverted? Are there any particular features in the recording that make the inversion obvious for you?

There might be other elements, but what was immediately apparent to me and what I listened for in the ABX test was the (possibly percussive) low bass sound within the first couple of seconds (maybe a hand or string striking the fretboard or guitar body?). This sound was more prominent in the inverted version, but also kind of more diffuse sounding - difficult to describe, but easy to hear. I'm curious, can anyone else hear this and successfully pass the ABX test I linked?

(Mods, feel free to move this to a new thread if it's deemed unrelated to the topic of this thread.)
 
Last edited:

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
There might be other elements, but what was immediately apparent to me and what I listened for in the ABX test was the (possibly percussive) low bass sound within the first couple of seconds (maybe a hand or string striking the fretboard or guitar body?). This sound was more prominent in the inverted version, but also kind of more diffuse sounding - difficult to describe, but easy to hear. I'm curious, can anyone else hear this and successfully pass the ABX test I linked?

Did you do the test on speakers or headphones?

(Mods, feel free to move this to a new thread if it's deemed unrelated to the topic of this thread.)

This is an interesting topic. There might be people that can actually hear absolute phase just like there are people that have absolute pitch.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,422
Likes
4,029
Location
Pacific Northwest
... Yes a polarity inversion is a 180º phase shift. One can usefully regard this as a neat coincidence rather than anything profound.
sin(wt + 180º) = -sin(wt) That is all.
...
That's often said but I'm not so sure. The equation is of course correct, but it refers to a sin wave which is a perfectly symmetric steady state oscillation. Shifting phase 180* slides it, and inverting polarity flips it. With a symmetric steady state wave like sin/cos, sliding it just the right amount happens to be the same as flipping it. But a musical waveform is neither symmetric nor steady state. For example, take a musical impulse like a stick hitting a drum. I can't imagine any amount of sliding it along X (time) axis that matches the waveform you get when flipping across Y (amplitude) axis.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,012
Location
Berlin, Germany
Polarity inversion is not the same as 180° phase shift. It is just that in the special case of a continuous (infinite) sine wave the waveform happens to look the same.
http://www.davidbridgen.com/pol.htm
 
Last edited:

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,012
Location
Berlin, Germany
This is an interesting topic. There might be people that can actually hear absolute phase just like there are people that have absolute pitch.
I have yet to meet a person that doesn't hear polarity inversion but there might be people who really don't preceive it no matter how hard they try.
The changes are exactly like @GaryH has described, the "body" (timbre) of bass low notes can change and the soundstage rendering often also changes. The reasons for the first effect are completey known and clear, for the latter much less so.
Same with considerable amounts of phase rotation. The most prominent case is the typical 80Hz 4th order acoustical XO of a properly dialed in Sub/Sat system, THX style.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,690
Likes
6,012
Location
Berlin, Germany
As for amp HF phase response, this is completely irrelevant.
People should be more concerned about the phase shift introduced by the analog post filters in their DACs, and of the DAC filters themselves if those are non linear phase.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,422
Likes
4,029
Location
Pacific Northwest
Nice example and link. To play devil's advocate, here's a thought: consider that asymmetric sawtooth wave example. Take the Fourier series of that sawtooth to get a set of cos() waves. Since the sawtooth has a discontinuous 1st derivative, the series will have an infinite # of terms/waves. Approximate it by truncating after a "large enough" number of waves. Now apply a 180* phase shift to each of those waves. This will "shift" low and high frequency waves by different amounts along the X axis. Now sum up all those phase shifted waves. What you get will not be the same as simply shifting the sawtooth by half its wavelength, because the low and high frequency components of the Fourier series shifted by different amounts.

Is it possible that the sum of all those individually shifted Fourier wave components actually is the same as inverting the polarity of the sawtooth? We could test this with a simple spreadsheet or Python code.
PS: I think this would indeed invert the polarity of the wave. Each individual Fourier component/wave would be shifted 180*, which inverts it. So it would be like inverting each of the individual cos() waves in the Fourier series. This is simply multiplication by -1, which can be factored out as -1 times the original Fourier series, which is the polarity inverted wave.
 
Last edited:

Robbo99999

Master Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
6,874
Likes
6,672
Location
UK
It is best to look at a few examples to get a sense for the phase shift.

Here would be a roll-off 1st order at 35kHz (FR and phase in green)
View attachment 134890

and here a roll-off 8th order at 35kHz (FR and phase in green)
View attachment 134891

The phase shift that would be caused by the amplifier roll-off is 15° at, for example, 10kHz with the first order filter and 80° with the eighth order filter.

However, the phase shift itself does not express a time delay. Only the group delay, as gradient of the phase curve, has a unit of time.
Even the group delay is usually not equivalent to a time delay (exceptions: pure time delay, allpass).
Source: Testing Loudspeakers, D'Appolito (quoted in substance)

In the examples it can be seen that the group delay is nearly 0, about 0.02ms for the eighth order filter, and is constant over the entire audible frequency range 20-20000Hz:
View attachment 134895
Even if one assumes that the increased group delay corresponds to a time delay, this would be constant and thus inaudible.
Or are there studies that come to a different conclusion?
.
I don't know, you tell me, I don't really understand phase on any deep level.
 
Top Bottom