• 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*)

such as the transfer function (s+w0)/(s+w0), where the parts each have infinite impulse response, but the sum of parts has a perfect impulse response.

One must avoid actual zeros in the magnitude response, of course. Not that yours is problematic. I say this because I have tested some systems that quite bluntly forgot about that.
 
Which may or may not be audible to some skeptics, but the converse, that it is NEVER AUDIBLE, is completely controverted by a simple matlab script. So we must discard the idea that it is NOT audible, so long, and goodbye. That claim is dead, dead, dead.
Totally agree on this.
 
Which may or may not be audible to some skeptics, but the converse, that it is NEVER AUDIBLE, is completely controverted by a simple matlab script. So we must discard the idea that it is NOT audible, so long, and goodbye. That claim is dead, dead, dead.
Well, it was always going to be obviously audible if the phase shift is so great that half of the frequencies are delayed by, say, one second. That's intuitively obvious and wouldn't even need a test signal demo.

The pragmatic conversation, IMHO, is with respect to the amounts of all-pass* phase shift and relative signal magnitudes that typically occur with audio gear and environments. Where is that audibly negative?

* (I use the term 'all-pass' phase shift as short hand for phase shift that isn't affecting the frequency response and consequently being audible because the change in FR is audible. If that is imprecise or plain wrong terminology it is because I am not well-versed in this area and its terms.)
 
Well, it was always going to be obviously audible if the phase shift is so great that half of the frequencies are delayed by, say, one second.

In such situations envelope delay is appropriate to consider, of course. It takes considerably less than that to make something audible. A half wave somewhere around 500Hz is nothing like that delay but it works just fine to demo the audibility of a tritone AM vs. FM.
 
Which may or may not be audible to some skeptics, but the converse, that it is NEVER AUDIBLE, is completely controverted by a simple matlab script. So we must discard the idea that it is NOT audible, so long, and goodbye. That claim is dead, dead, dead.

Motion to finally fix the thread title?
 
@theREALdotnet, The evidence was already there decades ago, for those who care. Blanket statements like this tend to be false most often anyway. In post #4 I proposed a test every sceptic could do for himself/herself.
I have little hopes the title correction is going to happen, but maybe I'm wrong this time, @amirm?
 
@theREALdotnet, The evidence was already there decades ago, for those who care. Blanket statements like this tend to be false most often anyway. In post #4 I proposed a test every sceptic could do for himself/herself.
I have little hopes the title correction is going to happen, but maybe I'm wrong this time, @amirm?
Judging by the speed people search about this stuff a google hit for this will present only the tittle,one will just see it and read nothing more.
It must absolutely change.
 
Last edited:
I have little hopes the title correction is going to happen, but maybe I'm wrong this time, @amirm?

We can't let Betteridge's Law win this time.
 
Clearly it has been shown that it is possible to create an example of audible phase shift. The question is does the normal level of phase shift found in a Hi-Fi audio system rise to the level of audibility? The only time I have "clearly" (reliably able to ABX) heard phase shift differences is when I made a mistake with DSP trying to correct phase and really screwed things up. The differences I have heard between a "normal non phase corrected system" and the same system phase corrected are subtle at best and very difficult to reliably ABX. Reading through this thread and seeing the difficulty of correctly measuring phase and understanding the measurements and corrections I am not sure more harm than good is often done. I have settled on using "FIR Linearized" IIR filters and figure it is close enough.
 
The question is does the normal level of phase shift found in a Hi-Fi audio system rise to the level of audibility?
For example, the phase shift from a subwoofer to mains crossover (typically an acoustic LR4 @ 80Hz if THX compliant) can be demonstrated to be audibly relevant to a number of people with a lot of music, in that they can "reliably" A/B its presence or absence (blind test of course, though not necessarily only ABX). It might need a lot of training and dedication/passion for the topic and the differences are generally subtle, but this doesn't make the results less valid. For me that's a 100% fact because I've conducted such tests, and so have others. For this and other phase-shift scenarios. including a simple polarity flip, just a special case of phase distortion.

Bottom line, if you really care, "normal levels of phase distortion" are likely to be audible to you.
And now let's move on....
 
I agree, the record needs to be corrected.

Clearly it has been shown that it is possible to create an example of audible phase shift. The question is does the normal level of phase shift found in a Hi-Fi audio system rise to the level of audibility? The only time I have "clearly" (reliably able to ABX) heard phase shift differences is when I made a mistake with DSP trying to correct phase and really screwed things up.

I have my own anecdote. I was aware of this thread and that experts think phase shift is inaudible. So one day a few months back, I decided to do a really ham fisted phase correction on my linear phase system. The verification measurement was the same as the control, except for the phase response which was commendably flat. So I sat down to have a listen. I was quite shocked to hear the soundstage had lifted up off the floor and the whole thing was vertically compressed, as if I was listening out of a letter box. It was not subtle at all, in fact it was quite horrible. I did not blind test, but I am pretty certain I could pick it 10 out of 10.

This lead me to months of investigation where I made and re-made filters with a lot of verification measurements. That made me realise that beyond a certain threshold, phase is indeed audible and if you deliberately throw out the rules of DSP and do something you shouldn't, awful results will follow. So the question now is, what is the threshold for audibility, and is there any point in pushing the correction further? Should we be content with phase distortion produced by all passive crossovers, all speakers with IIR DSP's, and in fact nearly every minimum phase speaker on the planet?

The new version of Acourate has a procedure for linearising the phase of each individual driver. I have to admit that I have some concerns about this, especially since I am taking the measurement from 1m away. @NTK has pointed out in another thread that in the nearfield, pressure and particle velocity are out of phase. So I may be taking an unreliable measurement, and therefore performing an unreliable correction. But I can report that it actually sounds quite good.

My own experiments suggest that there is indeed an advantage in going linear phase, but I am not going to convince anyone without evidence.
 
Clearly it has been shown that it is possible to create an example of audible phase shift. The question is does the normal level of phase shift found in a Hi-Fi audio system rise to the level of audibility?
That was my question in #1043.
The only time I have "clearly" (reliably able to ABX) heard phase shift differences is when I made a mistake with DSP trying to correct phase and really screwed things up.
...which was really a case of hearing the change in frequency response.
The differences I have heard between a "normal non phase corrected system" and the same system phase corrected are subtle at best and very difficult to reliably ABX.
Again I think we need to be clear that the example needs to be one where the frequency responses are identical. Otherwise we are just discussing the audibility of changes to the frequency response.
 
TAgain I think we need to be clear that the example needs to be one where the frequency responses are identical. Otherwise we are just discussing the audibility of changes to the frequency response.

The evidence from the tritone has exactly, precisely the same power spectrum, so there's that. No, it's not something we usually listen to.

If you want to hear effects at high frequencies, listen to the ring notes in "We Shall Be Happy" from Ry Cooder's "Jazz" album (first 30 seconds).
 
Last edited:
The evidence from the tritone has exactly, precisely the same power spectrum, so there's that. No, it's not something we usually listen to.
Yes, thank you, I understood that from your earlier posts, and I am sure you are getting tired of repeating it. ;)

If you want to hear effects at high frequencies, listen to the ring notes in "We Shall Be Happy" from Ry Cooder's "Jazz" album (first 30 seconds).
OK I will follow that up, thanks. It will be interesting.

Again, though, you are providing examples of signals that already have phase shift of a sort that is audible (with same power spectrum). Which is all very cool and interesting. Yes, let's take it as established that it is audible.

But what I am interested in also exploring, is this: is typical audio gear causing this type of phase shift (that doesn't affect the frequency response but is audibly not transparent), and causing it with sufficient magnitude to be audible? Is it common in loudspeakers? In amps? Are there specific speakers or amps that don't produce it at an audible level, or at least produce audibly less of it than other speakers/amps (and hence could be said to be audibly superior in this respect)?

Or, are we in reality not looking at an issue that we need to consider in our choice of hifi audio gear?

cheers
 

But what I am interested in also exploring, is this: is typical audio gear causing this type of phase shift (that doesn't affect the frequency response but is audibly not transparent), and causing it with sufficient magnitude to be audible? Is it common in loudspeakers? In amps? Are there specific speakers or amps that don't produce it at an audible level, or at least produce audibly less of it than other speakers/amps (and hence could be said to be audibly superior in this respect)?

Or, are we in reality not looking at an issue that we need to consider in our choice of hifi audio gear?

cheers

Speakers == yes.
 
Examples?
Most of the conventional multi-way speakers. The more ways and the lower the crossover frequencies the more prominent are the low-frequency effects of phase distortion (but there is impact at higher frequencies as well).

Extreme example would be a large 3-way with subs, so you'd have XO points at 80Hz, 300Hz, 3kHz ballpark. Listen to these as is vs with compensated excess phase and you'll have no more questions. LF-timbre is affected and sometimes one can hear the difference of the bass delay.

The phase distortion of small 2-way with a single 3kHz XO is quite benign in comparison but it may still be audibly when it is removed, most often it is the rendering of the soundstage that is different, phantom sources and diffuse reverb tails having different dimensions/sharpness, stuff like that.
 
Extreme example would be a large 3-way with subs, so you'd have XO points at 80Hz, 300Hz, 3kHz ballpark. Listen to these as is vs with compensated excess phase and you'll have no more questions. LF-timbre is affected and sometimes one can hear the difference of the bass delay.
Actually I would like to hear that.
 
Back
Top Bottom