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

[Hayabusa]

Continuing my reading about phase. I read David Griesinger's paper on proximity available here: http://www.ica2016.org.ar/ica2016proceedings/ica2016/ICA2016-0379.pdf

Summary of what he said: "proximity" is the sensation of being closer or further to a sound source. For example, if you listen to a string quartet and start walking further away, the sound changes from being able to distinguish individual musicians and being able to hear who played which note, to an indistinct fuzzy ball of sound. He says that phase alignment is what creates proximity, and loss of phase alignment causes loss of proximity. This loss of phase alignment is caused by reflections - the further you walk away, the worse it gets.

In home audio, I wonder if the equivalent would be: the minimum phase response of the speaker getting muddied up by the excess phase reflections of the room.

This also leads me to wonder whether loudspeakers that distort the phase response (i.e. any loudspeaker with a minimum phase crossover) would also reduce the "proximity" effect, and whether we should all be aiming for linear phase behaviour from loudspeakers. Opinions?

First impression would be that the phase variations you see from a speaker are way smaller than the variations caused by reflections, we are talking microseconds v.s milliseconds here..

 

[Thomas_A]

Perception of distance to sound source is dependent on loudness, high frequency content and DRR (direct to reflected sound ratio). Speaker phase response have little to do with this.

 

[gberchin]

What you want to do if you want to make your constant delay filter into a minimum delay filter is to use a cepstrum to convert the filter to its minimum phase version.

@j_j, I haven't had a chance to grok the Smith/Fergusen document, yet, but how does their method compare with that given is section 10.6.3 of the original Oppenheim and Schafer (1975)?

 

[Snarfie]

Continuing my reading about phase. I read David Griesinger's paper on proximity available here: http://www.ica2016.org.ar/ica2016proceedings/ica2016/ICA2016-0379.pdf

Summary of what he said: "proximity" is the sensation of being closer or further to a sound source. For example, if you listen to a string quartet and start walking further away, the sound changes from being able to distinguish individual musicians and being able to hear who played which note, to an indistinct fuzzy ball of sound. He says that phase alignment is what creates proximity, and loss of phase alignment causes loss of proximity. This loss of phase alignment is caused by reflections - the further you walk away, the worse it gets.

In home audio, I wonder if the equivalent would be: the minimum phase response of the speaker getting muddied up by the excess phase reflections of the room.

This also leads me to wonder whether loudspeakers that distort the phase response (i.e. any loudspeaker with a minimum phase crossover) would also reduce the "proximity" effect, and whether we should all be aiming for linear phase behaviour from loudspeakers. Opinions?when i bought time alignt/phase coherent speakers by build in combination with DSP an using a more or less near field listning distance using farfield colum speakers.

When i bought time alignt/phase coherent speakers by build in combination with DSP an using a more or less near field listening distance/Proximity (speaker 1,80 wide listening distance 2 meters see picture) using farfield colum speakers the whole audio image changed dramaticly for the better ( imaging, staging second to none imo) Curious was that i tested/ corrected several comparable colume speakers an all improved quite dramaticly for the better the only one that came on top was the Vandersteen time alignt phase coherent speakers by build. Most important difference between the Vandersteen an other colum speakers i tested where besides instrument especially voice sounds now like a panel/electrostatic speaker boxie sound is gone

If i put the speakers a bit wider an i sit between 2 a 3 meter meters away the whole audio picture collaps fun is gone.

 

[j_j]

Continuing my reading about phase. I read David Griesinger's paper on proximity available here: http://www.ica2016.org.ar/ica2016proceedings/ica2016/ICA2016-0379.pdf

Summary of what he said: "proximity" is the sensation of being closer or further to a sound source. For example, if you listen to a string quartet and start walking further away, the sound changes from being able to distinguish individual musicians and being able to hear who played which note, to an indistinct fuzzy ball of sound. He says that phase alignment is what creates proximity, and loss of phase alignment causes loss of proximity. This loss of phase alignment is caused by reflections - the further you walk away, the worse it gets.

That's part of it. I would phrase it as "there are 3 ranges" (which overlap and are not absolute, of course)

Very near - increase in interaural delays in HRTF - maybe up to 10 head "diameters". This can be incredibly striking, having done some gaming sound effects, like a weapon racking 2' behind the listener. (I'm not counting this in the 3 ranges, btw, for practical purposes.)

Near-ish (from somewhere above to a few head heights above ground) - Diffused floor reflections. (undifused floor reflections create "phasey" effects.)

Farther - Direct to diffuse ratio of the sound arrival.

Really farther than that: Diffusion of direct sound (first arrival) due to air movement.

In home audio, I wonder if the equivalent would be: the minimum phase response of the speaker getting muddied up by the excess phase reflections of the room.

Same issue as above, so yes, sometimes, no, sometimes, depends on the amount, time, and duration of the reflections. You betcha it's excess phase alright.

This also leads me to wonder whether loudspeakers that distort the phase response (i.e. any loudspeaker with a minimum phase crossover) would also reduce the "proximity" effect, and whether we should all be aiming for linear phase behaviour from loudspeakers. Opinions?

It's a different effect, but it's very real. It is usually completely overrun by crossover mismatch between channels, though. It's not a huge effect. More often the effective delay is different across the crossover point(s) due to the slight mismatch in components. It is very tricky to demonstrate without having something that is NOT suffering from the same problem to compare to. In our (pretty dead) listening room, with the AC off (to stop air movement) we can get a most startling 'in the head' effect just like headphones with our loudspeakers, which use IDENTICAL (to double precision float) constant-delay crossovers with driver compensation per driver set. That doesn't much happen with analog IIR crossovers at all.

You notice this mostly when you go back to a different set of speakers, and it's "whaaat the heck". I can't describe the difference as much as I can say you notice and it bothers you. I am pretty sure most of us adapt to our particular crossover setups for IIR crossovers that aren't digital,though.

 

[j_j]

@j_j, I haven't had a chance to grok the Smith/Fergusen document, yet, but how does their method compare with that given is section 10.6.3 of the original Oppenheim and Schafer (1975)?

Not having dug into the guts, all I can say is that sometimes one works better, sometimes the other works better, and neutralizing near-singular parts in the ceptstrum works more often. Of course, it's more of an approximation, but screwing up the minimum-phase nature for parts of the output that don't exist anyhow (around zeros) really doesn't matter much, does it?

 

[j_j]

Perception of distance to sound source is dependent on loudness, high frequency content and DRR (direct to reflected sound ratio). Speaker phase response have little to do with this.

Uh, I must disagree, but only in a carefully designed environment with a set of speakers that are constant delay, and do not have any differential phase response will you hear the comparison.

 

[gberchin]

.... screwing up the minimum-phase nature for parts of the output that don't exist anyhow (around zeros) really doesn't matter much, does it?

As long as those zeroes remain inside the unit circle.

 

[j_j]

As long as those zeroes remain inside the unit circle.

Ya think?

 

[Thomas_A]

Uh, I must disagree, but only in a carefully designed environment with a set of speakers that are constant delay, and do not have any differential phase response will you hear the comparison.

You mean a traditional LR filter at 3 kHz between mid and tweeter has any meaning for distance perception? Where is that publication?

 

[j_j]

You mean a traditional LR filter at 3 kHz between mid and tweeter has any meaning for distance perception? Where is that publication?

You did read the rest of the discussion, yes? No, there is no paper. Nobody pays me to write papers any more.

 

[Thomas_A]

You did read the rest of the discussion, yes? No, there is no paper. Nobody pays me to write papers any more.

But there are others that do research. Is it audible through headphones?

 

[theREALdotnet]

In home audio, I wonder if the equivalent would be: the minimum phase response of the speaker getting muddied up by the excess phase reflections of the room.

This also leads me to wonder whether loudspeakers that distort the phase response (i.e. any loudspeaker with a minimum phase crossover) would also reduce the "proximity" effect, and whether we should all be aiming for linear phase behaviour from loudspeakers. Opinions?

In my opinion and limited experience, yes and yes.

Is it time for @amirm to fix the thread title?

 

[j_j]

But there are others that do research. Is it audible through headphones?

Is WHAT audible? When we are taking about speaker crossovers, how are headphones involved? Perhaps you could phrase a question more clearly?

Also, FYI I still do research, I just no longer get paid to give it away, mm'kay?

 

[Thomas_A]

Is WHAT audible? When we are taking about speaker crossovers, how are headphones involved? Perhaps you could phrase a question more clearly?

Also, FYI I still do research, I just no longer get paid to give it away, mm'kay?

A phase shift caused by a proper LR filter at 3 kHz. The question was related to proximity perception.

 

[j_j]

A phase shift caused by a proper LR filter at 3 kHz. The question was related to proximity perception.

From what position relative to the radius to the ?2? drivers. Or are we talking wmtmw? or just mtm? What order LR? What signal? How much energy in the signal around the crossover range that does not have a flat envelope? In general an LR crossover does have an allpass characteristic, yes? Does it exceed about about pi/8 phase shift inside of any ERB?

While it may not apply in practice, if you give me the phase response of the LR crossover, I can probably design a signal that's audible. Maybe not for very low order, effectively providing no tweeter protection.

Let's also talk in that case about the reflected sound in the listening room, although that's not directly phase related.

 

[Thomas_A]

From what position relative to the radius to the ?2? drivers. Or are we talking wmtmw? or just mtm? What order LR? What signal? How much energy in the signal around the crossover range that does not have a flat envelope? In general an LR crossover does have an allpass characteristic, yes? Does it exceed about about pi/8 phase shift inside of any ERB?

While it may not apply in practice, if you give me the phase response of the LR crossover, I can probably design a signal that's audible. Maybe not for very low order, effectively providing no tweeter protection.

Let's also talk in that case about the reflected sound in the listening room, although that's not directly phase related.

A fourth order LR at 3 kHz, 5-6 inch mid woofer and dome tweeter, distance 2-3 meters, music. Again related to proximity perception of the crossover phase shift.

 

[Keith_W]

That's part of it. I would phrase it as "there are 3 ranges" (which overlap and are not absolute, of course)

Very near - increase in interaural delays in HRTF - maybe up to 10 head "diameters". This can be incredibly striking, having done some gaming sound effects, like a weapon racking 2' behind the listener. (I'm not counting this in the 3 ranges, btw, for practical purposes.)

Near-ish (from somewhere above to a few head heights above ground) - Diffused floor reflections. (undifused floor reflections create "phasey" effects.)

Farther - Direct to diffuse ratio of the sound arrival.

Really farther than that: Diffusion of direct sound (first arrival) due to air movement.

That is fascinating. I don't play games, but now that you mention it, I did try my friend's gaming rig with headphones. The game was able to make gunshots sound further away. I thought it was due to volume, but are you saying that some games deliberately "fuzzy up" the phase to make them sound further away?

I learnt in art classes when we were studying body proportions that the body is between 6 to 8 heads tall, and the width of the head is about 2/3 the height of the head. We learnt to manipulate these proportions depending on what we wanted to convey about the character, but that is another story. The average head height is 26-28cm (10-11"), and the average width is about 18cm (7"). So, expressing your comment above with knowledge of proportions:

Very near = ("10 head diameters") = about 1.8m (6ft)
Near-ish = if we take "few head heights above the ground" to be 28cm (head height) x 7 (above the ground) x 3 ("a few") = about 5.9m (19 feet)

I listen to my stereo about 3.5m away, and the system is spaced away from the side walls so that the floor bounce is much earlier than the sidewall reflection. I suppose that I am getting mostly diffuse floor and ceiling reflections.

It's a different effect, but it's very real. It is usually completely overrun by crossover mismatch between channels, though. It's not a huge effect. More often the effective delay is different across the crossover point(s) due to the slight mismatch in components. It is very tricky to demonstrate without having something that is NOT suffering from the same problem to compare to. In our (pretty dead) listening room, with the AC off (to stop air movement) we can get a most startling 'in the head' effect just like headphones with our loudspeakers, which use IDENTICAL (to double precision float) constant-delay crossovers with driver compensation per driver set. That doesn't much happen with analog IIR crossovers at all.

You notice this mostly when you go back to a different set of speakers, and it's "whaaat the heck". I can't describe the difference as much as I can say you notice and it bothers you. I am pretty sure most of us adapt to our particular crossover setups for IIR crossovers that aren't digital,though.

Some time ago I performed an experiment with Acourate to linearise the phase of all my drivers, effectively converting each driver from minimum phase to linear phase. I took the measured phase, time reversed it, and used that as my preconditioning filter. Measurements confirmed that the phase was dead flat (see graph below; red = before, green = after). The verification measurement was a beautiful flat line, just look at it. I then time aligned all the drivers and the result was linear phase from 200Hz up. But ... the listening result was catastrophically bad. It seemed to lift the soundstage off the floor and compress it vertically, as if I was listening through a slit in a pillbox. No blind test needed. It was, as you say, a "whaaaat the heck" moment.

Is it time for @amirm to fix the thread title?

Considering that amirm's video caused me a lot of confusion, I would say it's time for a new video. You can really screw phase up with DSP if you are not paying attention, and given that I thought that phase was inaudible (thanks to that video) I did not pay attention. The result was a lot of very strange sounding filters, which prompted my investigation into this phenomenon. Having spent a lot of time investigating this and making dozens of filters, I would confidently say that phase effects are easily audible if they are large enough.

I don't think it was wasted time though, I learnt a lot in this process.

 

[j_j]

A fourth order LR at 3 kHz, 5-6 inch mid woofer and dome tweeter, distance 2-3 meters, music. Again related to proximity perception of the crossover phase shift.

That, specifically, requires some math exercise. If you happen to have the allpass phase handy, you could tell me. For this I assume you mean run the crossover on the input signal, add the two outputs, and see if the phase shift is audible in headphones? As a very rough guess I would say it's on the edge, and sensitivity to analog component issues in an analog crossover would be an issue. In a digital version, the power vs. direct response would not be a phase issue, but a timbre issue.

 

[Thomas_A]

That, specifically, requires some math exercise. If you happen to have the allpass phase handy, you could tell me. For this I assume you mean run the crossover on the input signal, add the two outputs, and see if the phase shift is audible in headphones? As a very rough guess I would say it's on the edge, and sensitivity to analog component issues in an analog crossover would be an issue. In a digital version, the power vs. direct response would not be a phase issue, but a timbre issue.

Regarding timbre shift, I have preciously shown my own positive DBT rewults which I hear when all-passed at lower frequencies. Some people seem not to hear it though. But above 1 kHz. I can’t hear it.