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

[DaneelOlivaw]

It was shown here that the phase change of the amplifier is not relevant for sound perception. But the reproductive accuracy of the phase difference between left and right can affect spatial perception. And this can depend on the bandwidth of the amplifier. Of course, you also need time-aligned speakers for accuracy.

 

[Holmz]

It was shown here that the phase change of the amplifier is not relevant for sound perception. But the reproductive accuracy of the phase difference between left and right can affect spatial perception. And this can depend on the bandwidth of the amplifier. Of course, you also need time-aligned speakers for accuracy.

^that’s^ sort of a problem, when the rating are for the radiation pattern, and the flatness of the sound versus freqnecy.

 

[j_j]

It was shown here that the phase change of the amplifier is not relevant for sound perception. But the reproductive accuracy of the phase difference between left and right can affect spatial perception. And this can depend on the bandwidth of the amplifier. Of course, you also need time-aligned speakers for accuracy.

Which amplifier? Are we talking about pure delay, which can easily be a million degrees at 20khz and not be important at all.

Now, if it's a lot of phase shift after delay is removed, OR if it's dependent on the frequency of another signal in the same amplifier, then it may be questionable. I have met a FEW amplifiers that had phase shift that varied slightly with bass frequencies, for some mechanism I don't understand. Fortunately it wasn't enough to cause noticeable FM modulation in that particular case.

And, of course, an amplifier that slew rate limits can be a sonic catastrophe if it's bad enough, but I think those are gone now. I hope they are.

 

[board]

Which amplifier? Are we talking about pure delay, which can easily be a million degrees at 20khz and not be important at all.

Now, if it's a lot of phase shift after delay is removed, OR if it's dependent on the frequency of another signal in the same amplifier, then it may be questionable. I have met a FEW amplifiers that had phase shift that varied slightly with bass frequencies, for some mechanism I don't understand. Fortunately it wasn't enough to cause noticeable FM modulation in that particular case.

And, of course, an amplifier that slew rate limits can be a sonic catastrophe if it's bad enough, but I think those are gone now. I hope they are.

I just want to make sure I understand what you mean, as I'm much more of a novice in audio than you are.
Would it be fair to say that you mean that in almost all amplifiers on the market today there is no phase shift or phase differences between the two stereo channels that would be audible?
And then there might be a few very specific amplifiers with broken designs where it could possibly be audible?

 

[j_j]

I just want to make sure I understand what you mean, as I'm much more of a novice in audio than you are.
Would it be fair to say that you mean that in almost all amplifiers on the market today there is no phase shift or phase differences between the two stereo channels that would be audible?
And then there might be a few very specific amplifiers with broken designs where it could possibly be audible?

Well, yes and no. There are some amplifiers that have issues near overload (some switching "class D" amplifiers). There are some "analog" amplifiers that can't handle some specific kinds of load. In general, a well-designed amplifier of any kind is not likely to create much audible distortion and noise, it's a several-times solved problem.

On the other hand, some receivers, etc, have issues with cheaping out on output devices, shared power supplies, and the like. In terms of phase shift, beyond delay that is part of the nature of switching amps, phase shift alone should not generally be a problem.

I will never, however, assert that everyone 'gets it right' with a design.

As to the slew rate limiting issues, or the signal-dependent issues, that's USUALLY a second-order effect. Again, execution is the question. Most modern amplifiers should be pretty good in applications they are intended for, but there are exceptions, especially in the ultra-cheap and boutique categories.

 

[welwynnick]

Would it be fair to say that you mean that in almost all amplifiers on the market today there is no phase shift or phase differences between the two stereo channels that would be audible?
And then there might be a few very specific amplifiers with broken designs where it could possibly be audible?

I'm late to the party, but have been keen to look at the objective and subjective aspects of phase variation in each piece of audio equipment.
Speakers and the room are the worst offenders. DACs are another matter for another day, but I did wonder about amps.
There's not much published information, but Soundstage have started doing phase measurements of preamps and power amps.
In general they all have very little phase variation, maybe a few tens of degrees at the top end at worst, so nothing to worry about.
However there's always an exception to prove the rule, and I found the Peachtree Audio Carina GaN Integrated Amplifier, which was rather different:

SoundStageNetwork.com | SoundStage.com - Peachtree Audio Carina GaN Integrated Amplifier Measurements

www.soundstagenetwork.com

 

[René - Acculution.com]

I'm late to the party, but have been keen to look at the objective and subjective aspects of phase variation in each piece of audio equipment.
Speakers and the room are the worst offenders. DACs are another matter for another day, but I did wonder about amps.
There's not much published information, but Soundstage have started doing phase measurements of preamps and power amps.
In general they all have very little phase variation, maybe a few tens of degrees at the top end at worst, so nothing to worry about.
However there's always an exception to prove the rule, and I found the Peachtree Audio Carina GaN Integrated Amplifier, which was rather different:

SoundStageNetwork.com | SoundStage.com - Peachtree Audio Carina GaN Integrated Amplifier Measurements

www.soundstagenetwork.com

View attachment 466952

This is likely just a pure time delay, as they also mention. They should have subtracted the corresponding phase to see what is left phase-wise.

 

[gberchin]

This is likely just a pure time delay

Had the phase characteristic been plotted on a linear frequency scale, this would be obvious.

 

[René - Acculution.com]

Had the phase characteristic been plotted on a linear frequency scale, this would be obvious.

Yes exactly, and a straight line could be overlaid to better see any difference between the two. Oh well.

 

[j_j]

The first thing to do is to figure out the linear regression in LINEAR FREQUENCY (not log frequency) of that phase shift. Extract the linear part, because that's time delay. Then see how much actual phase shift there is beyond pure time delay.

 

[witwald]

The first thing to do is to figure out the linear regression in LINEAR FREQUENCY (not log frequency) of that phase shift. Extract the linear part, because that's time delay. Then see how much actual phase shift there is beyond pure time delay.

Wouldn't that process first require unwrapping the phase response? I can't see how a linear regression might otherwise be performed on the existing phase data, which has many multiples of 360° of phase shift in it. If I recall correctly, programs such as MATLAB have functions that do just that.

 

[NTK]

Wouldn't that process first require unwrapping the phase response? I can't see how a linear regression might otherwise be performed on the existing phase data, which has many multiples of 360° of phase shift in it. If I recall correctly, programs such as MATLAB have functions that do just that.

Yes. Unwrap phase first. J J has explained the procedure before.

...
You apparently have access to matlab and some source of head related impulse responses. Take a great big, long, FFT of that impulse response. Calculate the phase of the positive frequencies using atan2. Then "unwrap" the phase. Do a first order linear fit (polyfit) to the phase. Subtract out the part amounting to the linear fit from the phase. The slope of that linear fit, by the way, is exactly the overall time delay. Take the difference. In real data I've seen, that will not be a straight line. The variation from the straight line can be directly converted to +- time of arrival across frequency, remember phi = 2 * pi * f *t You have phi and f, now calculate 't'.

 

[amirm]

What does an impulse response have to do with being representative of music? It's simply a measure of system response.

I explained. An impulse has infinite bandwidth and energy. Here is its FFT:

As such, it is an "illegal" signal, certainly in the context of bandwidth limited digital music. It also has infinite rise time, again an illegal real life signal.

An impulse is a conceptual signal that is useful in engineering analysis but you don't want to confuse yourself into thinking your system needs to reproduce it with high fidelity. Such an attempt will lead to designs that have very wide bandwidth but otherwise, have poor response (former doesn't come for free).

 

[Newman]

I explained. An impulse has infinite bandwidth and energy. Here is its FFT:

As such, it is an "illegal" signal, certainly in the context of bandwidth limited digital music. It also has infinite rise time, again an illegal real life signal.

An impulse is a conceptual signal that is useful in engineering analysis but you don't want to confuse yourself into thinking your system needs to reproduce it with high fidelity. Such an attempt will lead to designs that have very wide bandwidth but otherwise, have poor response (former doesn't come for free).

FYI, Gnarly and I have had this conversation before, BTW, actually concerning square wave reproduction, but the same issue.

See half-way down each post, where square waves are mentioned. Post 1. Post 2. I hope you are more successful than I: he called my explanation "technobabble". I know I am not the best of communicators, but I try.

cheers

 

[pma]

Misleading info has been posted about impulse response here.

Impulse response of a linear system is an inverse Fourier Transform of Frequency Response. It does not need to be measured with an ideal Dirac impulse. REW or Arta calculate impulse response from frequency response. Step response is a derivative of impulse response. Please guys keep the basics of maths and physics without modifications.

 

[Holmz]

I explained. An impulse has infinite bandwidth and energy. Here is its FFT:

…

As such, it is an "illegal" signal, certainly in the context of bandwidth limited digital music. It also has infinite rise time, again an illegal real life signal.

Is a single bit error, or all zeros with a 1 or all/somebits being non-zero also illegal?
It seems like a bit error in a digital file, or a pop or click on an LP would also then be illegal… but as they seem to happen, they then must be a different law than the laws of physics and engineering.

It also does not extend all the way to DC, in addition to extending to infinity.

An impulse is a conceptual signal that is useful in engineering analysis but you don't want to confuse yourself into thinking your system needs to reproduce it with high fidelity. Such an attempt will lead to designs that have very wide bandwidth but otherwise, have poor response (former doesn't come for free).

I think that the idea of describing how the system perturbs the phase is what is being gleaned from the step function response.

But the question for the luminaries is whether perturbations of the phase in the direct sound, off of the speaker, is important at all... And whether there is much in teh way of ABX or other studies that say it has even one half of an iota of relevance?

 

[Holmz]

Misleading info has been posted about impulse response here.

Impulse response of a linear system is an inverse Fourier Transform of Frequency Response. It does not need to be measured with an ideal Dirac impulse. REW or Arta calculate impulse response from frequency response. Step response is a derivative of impulse response. Please guys keep the basics of maths and physics without modifications.

^That^ sir is illegal!
Hence I am going to refer to you as “Felonious Monk” from now on.
And also call for the defenestration of ASR in Prague.

 

[pma]

And also call for the defenestration of ASR in Prague.

Fom a window of the Old Royal Palace.

Defenestrations of Prague - Wikipedia

en.wikipedia.org

 

[Holmz]

From a window of the Old Royal Palace.

Defenestrations of Prague - Wikipedia

en.wikipedia.org

Yeah @pma Prague is a fantastic city to visit.
We came in through Domelici, then Pilsen and onto Prague.
Then we stayed in Salsava (about 50km to the East), and then down towards Ceski Budawick, and to Linz.
I probably should’ve picked up some tonearm wire there in Linz… but we went to KZ Mathausen instead.

We were on a tandem bicycle, which is a great test of a relationship, unless the relationship is fragile and one actually wants the relationship to last.

That was in ‘99, and the Danube was pretty much near flood levels so we sort of winged it on what to do after scratching off just going down the Danube toward Yugoslavia. Maybe it was Croatia by then… It sort of all imploded after Tito.

At least with a Tandem everyone’s feets are synced together, although it is not uncommon to phase the rear peddles a few degrees out from the front peddles…

 

[Sokel]

Yeah @pma Prague is a fantastic city to visit.
We came in through Domelici, then Pilsen and onto Prague.
Then we stayed in Salsava (about 50km to the East), and then down towards Ceski Budawick, and to Linz.
I probably should’ve picked up some tonearm wire there in Linz… but we went to KZ Mathausen instead.

We were on a tandem bicycle, which is a great test of a relationship, unless the relationship is fragile and one actually wants the relationship to last.

That was in ‘99, and the Danube was pretty much near flood levels so we sort of winged it on what to do after scratching off just going down the Danube toward Yugoslavia. Maybe it was Croatia by then… It sort of all imploded after Tito.

At least with a Tandem everyone’s feets are synced together, although it is not uncommon to phase the rear peddles a few degrees out from the front peddles…

The new Vltava Philharmonic Hall is of my very near future visits, can't wait for it.

Edit: I now saw it starts at 2033, not so near