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

[anphex]

Short question and I don't mind a short harsh answer:
I noticed a difference between my SMSL SU-9N Hybrid and Fast Linear Filter. Fast Linear sounds (yes, sorry for those words) more precise, less muddy.

Why is that? I concluded that any Ringing is inaudible. Phase shift should also be inaudible. The wider bandwith from 20 Hz to 18 Khz vs. 20 Hz to 22 Khz can't be it, right?

 

[DavidMcRoy]

Short question and I don't mind a short harsh answer:
I noticed a difference between my SMSL SU-9N Hybrid and Fast Linear Filter. Fast Linear sounds (yes, sorry for those words) more precise, less muddy.

Why is that? I concluded that any Ringing is inaudible. Phase shift should also be inaudible. The wider bandwith from 20 Hz to 18 Khz vs. 20 Hz to 22 Khz can't be it, right?

Slightly off topic, but I find that correctly implemented 4th-order Bessel filters modeled in DSP for crossover networks sound pristine clean compared with Butterworth and Linkwitz-Riley. I wonder if any double-blind testing has ever established that? My experience is only sighted. It's my understanding that they display less ringing. How audible is that, and could that be the effect I'm hearing? (It's one of those things that are subtle, yet "significant" if you're listening for it.)

 

[Holmz]

Short question and I don't mind a short harsh answer:
I noticed a difference between my SMSL SU-9N Hybrid and Fast Linear Filter. Fast Linear sounds (yes, sorry for those words) more precise, less muddy.

Why is that? I concluded that any Ringing is inaudible. Phase shift should also be inaudible. The wider bandwith from 20 Hz to 18 Khz vs. 20 Hz to 22 Khz can't be it, right?

If one could compare the two mathematically there is likely a difference.

 

[tomeh]

Well, to be clear, the pressure output of a loudspeaker is not minimum phase in general, although it is a physical system. The crossover alone ensures that, as it sums several transfer functions, and although they may be minimum phase on their own, their sum is not in general min ph (as the crossover typically aims to be an allpass filter when summed). Something like a step response will look very distorted and clearly shows the temporal issue, but we are lucky enough that we are not too sensitive to that. I will show a lot of this in the upcoming presentation. I am just afraid that people will not like a 2 hour lecture on this, as I have seen complaints about much shorter videos. But there is just no way around getting in to complex numbers and transfer functions to properly explain the many aspects of phase.

And would the sum of many steps/stages not create and unpredictable result? COuld you not daisy chain many filters to simulate it and then listen for perceived differences?

 

[j_j]

There are many issues in loudspeaker crossovers. That's a long discussion in and of itself, relating not only to summation, but also time delay, resulting changes from crossover phase shift with angle relative to drivers, and a bunch other stuff.

 

[Holmz]

]

Well, to be clear, the pressure output of a loudspeaker is not minimum phase in general, although it is a physical system. The crossover alone ensures that, as it sums several transfer functions, and although they may be minimum phase on their own, their sum is not in general min ph (as the crossover typically aims to be an allpass filter when summed). Something like a step response will look very distorted and clearly shows the temporal issue, but we are lucky enough that we are not too sensitive to that. I will show a lot of this in the upcoming presentation. I am just afraid that people will not like a 2 hour lecture on this, as I have seen complaints about much shorter videos. But there is just no way around getting in to complex numbers and transfer functions to properly explain the many aspects of phase.

Don’t they use a term “fck-em” in Europe?

Your last video was a dandy, and if the smart fellows and women go through it faster than me, then good for them.

But if it is similar to the last one in complexity, then based upon what I saw, I would suspect it is going to be long enough to get complaints from the smart people. But you may need it to be a similar length to account for those of us whom are a bit slower?

 

[Robbo99999]

There are many issues in loudspeaker crossovers. That's a long discussion in and of itself, relating not only to summation, but also time delay, resulting changes from crossover phase shift with angle relative to drivers, and a bunch other stuff.

I've not been following this whole discussion, but if there's a lot of stuff to consider with the "relatively controlled" notion of crossovers in a speaker, then where does this leave us when implementing seperately and somewhat imprecisely located subwoofers....even more variables there right? I recently bought my first sub, a worthwhile addition I think but it's not perfect. I measured it with a tape measure in it's physical distance vs the other 2 speakers in it's straight line to the listening position, and accounted for that with time delays in a miniDSP 2x4, measured it with a mic in REW to determine the crossover position and manipulated phase in the subwoofer (adjustable from 0 through to 180 degrees in 1 degree increments) to have the largest bass response at & around the crossover, then did roomEQ on the bass using REW and the miniDSP - I still can't help thinking this is still a much more uncontrolled situation than the nuances of crossovers in speakers that you're all talking about? Where does it leave people with what seems like the relatively uncontrolled nature of implementing subwoofers?

 

[Holmz]

… Further, do we measure pressure, velocity, or all 4 variables?

(I'm well aware you know about this, but it is truly discouraging how many people don't consider more than just pressure.)

Can you please outline the 4 variables?

Velocity is a function of temperature, so I am not sure if we need 1, 2, or 3 variables?

 

[j_j]

Can you please outline the 4 variables?

Velocity is a function of temperature, so I am not sure if we need 1, 2, or 3 variables?

Seriously? XYZW is a standard format, for example. Velocity means "volume velocity" and consists of 3 variables in a free field. They can be R theta phi, x, y, z, or whatever, plus of course the pressure component.

 

[j_j]

I've not been following this whole discussion, but if there's a lot of stuff to consider with the "relatively controlled" notion of crossovers in a speaker, then where does this leave us when implementing seperately and somewhat imprecisely located subwoofers....even more variables there right? I recently bought my first sub, a worthwhile addition I think but it's not perfect. I measured it with a tape measure in it's physical distance vs the other 2 speakers in it's straight line to the listening position, and accounted for that with time delays in a miniDSP 2x4, measured it with a mic in REW to determine the crossover position and manipulated phase in the subwoofer (adjustable from 0 through to 180 degrees in 1 degree increments) to have the largest bass response at & around the crossover, then did roomEQ on the bass using REW and the miniDSP - I still can't help thinking this is still a much more uncontrolled situation than the nuances of crossovers in speakers that you're all talking about? Where does it leave people with what seems like the relatively uncontrolled nature of implementing subwoofers?

Well, above 40Hz, subwoofers (unless you have at least 3) will not give you the proper physical sensation. We prefer to have full range "channels" instead.
The question of crossovers is secondary to that in my opinion.

 

[Robbo99999]

Well, above 40Hz, subwoofers (unless you have at least 3) will not give you the proper physical sensation. We prefer to have full range "channels" instead.
The question of crossovers is secondary to that in my opinion.

Hmm, ok, it seems like a good experience with just the one subwoofer, although I could do with more to help even out room modes and to reduce distortion (subwoofer not working as hard)........just it occurred to me there could be some parallels re what you folks are talking about and the "unknowns & innaccuracies" of incorporating/crossing over subwoofers with speakers.

 

[GaryH]

Well, above 40Hz, subwoofers (unless you have at least 3) will not give you the proper physical sensation.

Interesting, why 3?

 

[DonH56]

Slightly off topic, but I find that correctly implemented 4th-order Bessel filters modeled in DSP for crossover networks sound pristine clean compared with Butterworth and Linkwitz-Riley. I wonder if any double-blind testing has ever established that? My experience is only sighted. It's my understanding that they display less ringing. How audible is that, and could that be the effect I'm hearing? (It's one of those things that are subtle, yet "significant" if you're listening for it.)

Technically Bessel filters are linear-phase with constant group delay, at least ideally, so have better pulse response albeit slower roll-off than Butterworth. Linkwitz-Riley is a cascaded Butterworth topology that better matches amplitude through the crossover region (to avoid a hump or dip). In the world of radar and such, Bessel (and other filters, e.g. Chebschev and elliptical, though they have amplitude ripple the Bessel does not) are used more than Butterworth (etc.) for their superior pulse response, but in the audio world I am struggling to see how much it matters given all the other factors at play. I suspect it is more a matter of integration having to do with amplitude and phase alignment than the actual filter type or (im)pulse response.

Many years ago when I was fooling around with a crossover for my sub and Maggies (two crossovers, sub and bass/mid+tweeter) I implemented a Bessel filter, found it rolled off too slowly for reasonable order (this was before DSP so all analog RLC filters), tried a hybrid Bessel/Butterworth, briefly flirted with Chebyshev and elliptical, and ultimately implemented a third-order Butterworth. They all sounded different but it was all due to how the transition band was configured and tuned. These days, DSP makes it fairly easy, but back then I was adjusting time constants and tweaking all-pass stages (and even physical driver placement) to align everything. After careful tweaking any filter pretty much sounded like any other so I ended up with a standard Butterworth for ease of design and tweaking. This was before I knew of the Linkwitz-Riley approach that probably would have saved me some grief.

FWIWFM - Don

 

[pma]

Basically, type of the crossover filter transfer function and order of the filter are most important for directivity plots, together with driver axial distances and height difference. There would be a lot to speak about, however better to get the literature.

2nd order crossover, point source radiators, at crossover frequency.
3kHz crossover frequency, 14cm distance between point sources.

 

[fpitas]

https://www.richardfarrar.com/are-your-speakers-wired-correctly/

This page has a Speaker Phase Test.
The difference between "In phase" and "Out of phase" is audible.

What you're describing is essentially a crossover. When the sound waves from multiple drivers reinforce, or cancel, the resulting frequency response is affected.

 

[j_j]

Basically, type of the crossover filter transfer function and order of the filter are most important for directivity plots, together with driver axial distances and height difference. There would be a lot to speak about, however better to get the literature.

View attachment 236987
2nd order crossover, point source radiators, at crossover frequency.
3kHz crossover frequency, 14cm distance between point sources.

Indeed that distance makes things "extremely interesting".

 

[René - Acculution.com]

]


Don’t they use a term “fck-em” in Europe?

Your last video was a dandy, and if the smart fellows and women go through it faster than me, then good for them.

But if it is similar to the last one in complexity, then based upon what I saw, I would suspect it is going to be long enough to get complaints from the smart people. But you may need it to be a similar length to account for those of us whom are a bit slower?

I will make a shorter video on the topic when Erin’s Audio Corner is up for it. Any questions that anyone has on this here forum are most welcome. Ask away.

 

[Holmz]

Seriously? XYZW is a standard format, for example. Velocity means "volume velocity" and consists of 3 variables in a free field. They can be R theta phi, x, y, z, or whatever, plus of course the pressure component.

The speakers were more modelled as point source drivers with a distance separating them.
And added together in to create a superposition of the drivers summed up.

At sub frequencies there is no XYZ in the same way that a one does not hide behind a door in a bomb blast. The pressure only falls off with R^2.

Or maybe a what is volumes velocity? I get the vector part, but what volume?
Is it like wind? (I thought it was a pressure field?)

 

[j_j]

The speakers were more modelled as point source drivers with a distance separating them.
And added together in to create a superposition of the drivers summed up.

At sub frequencies there is no XYZ in the same way that a one does not hide behind a door in a bomb blast. The pressure only falls off with R^2.

Or maybe a what is volumes velocity? I get the vector part, but what volume?
Is it like wind? (I thought it was a pressure field?)

There is most certainly a volume velocity field in any soundfield. I've no idea what you're confused about, but you need to look at the basics of sound propagation, ok?

 

[Holmz]

There is most certainly a volume velocity field in any soundfield. I've no idea what you're confused about, but you need to look at the basics of sound propagation, ok?

I was hoping that on a forum you could explain it.
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.

The pressure field may have a vector describing the direction, which is radial for a point source… and it beams for a larger driver at a high frequency.

Where is the volume? Is it a Spatial volume? Or is it a Sound volume?