Minimum Phase, constant delay (aka linear phase), whazzit?

[levimax]

Absolutely. First, fix the acoustics. Part of the problem with room correction aside from the positional aspect was discussed in the subwoofer thread, in that big problems in room response that create zeros (dips) mean there is energy STORAGE in the room. Adding more to that storage is, err, problematic in most cases.

Can you make the room acoustics analogy that "fixing the acoustics of too small a sub enclosure by making it bigger is better accoustcally than using DSP" ?

 

[j_j]

Can you make the room acoustics analogy that "fixing the acoustics of too small a sub enclosure by making it bigger is better accoustcally than using DSP" ?

I could, but I might be making a bad analogy. A too small sub enclosure can't be linear, and you'd need positional feedback to make it work. Yes, some subs of that sort do exist.

Consider. If the cone can change the interior volume of the enclosure by 10%, that changes the spring constant the cone is working against by 10%, etc.

 

[gberchin]

Until you get to modern video codecs that have 5 or 6 frame minimum latency and memory.

Sure, but in my living room I am at the mercy of the audio/video sync provided by the cable company, or the streaming service, or etc. And they tend to synchronize their audio with their video. So if my audio processing adds additional delay beyond a few tens of milliseconds, I get lipsync problems.

 

[gberchin]

... big problems in room response that create zeros (dips) mean there is energy STORAGE in the room. Adding more to that storage is, err, problematic in most cases.

Interestingly enough, I just implemented an experimental Double Bass Array in my listening room. In theory, it absorbs some of the bass energy. Well, I'm not certain that "absorbs" is quite the appropriate term. But the measured results in my room are very encouraging. In short: it works.

 

[j_j]

Interestingly enough, I just implemented an experimental Double Bass Array in my listening room. In theory, it absorbs some of the bass energy. Well, I'm not certain that "absorbs" is quite the appropriate term. But the measured results in my room are very encouraging. In short: it works.

It's not impossible, but it has to be done very precisely.

 

[danadam]

modulo the annoying scaling issues with Matlab plotting.

sorry for the squished plot, but the autoscaling is a pain to fsck with

Any reason why not to set the axes manually?

 

[DWPress]

I've played with this by creating the exact same filter sets in FIR and IIR via plugins and alternating with another plugin A/B switch. Sited listening I can't hear any difference but I know my filters are good to start with. To make it even more interesting I create my filters for an active XO 3-way speaker + 2 subs (8ch) via REW and rePhase but hosted by Mitchco's Hang Loose Convolver which allows me to switch with zero latency. For kicks I created the same filters in rePhase using min, lin and mixed phase filters for the XO which is very audible across the whole frequency band and all my corrections are below 350Hz.

I haven't done any proper measurements to determine phase issues interacting with the room and the FIR vs IIR filter sets yet. A little more room treatment behind the speakers and I'll dive into a more thorough set of measurements.

If you're creating filters tight enough for only one place in time and space, you might as well switch to headphones.

 

[j_j]

Any reason why not to set the axes manually?

Laziness.

Well, and having to adapt to a fixed-ratio box for the data, too. Since I can expand and stretch the plot box...

 

[Lars Risbo]

Oh, and how you do all this follows in Matlab script:

clc
clear all
close all

len=64;

bb=firpm(len-1,[0 .35 .5 1],[1 1 0 0]);

freqz(bb)


bbr=roots(bb);

t=bbr;

for ii=1:len-1
if abs(bbr(ii)) > 1.0
bbr(ii)=1/bbr(ii);
end
end

cc=poly(bbr);
cc=cc/sum(cc);
figure
freqz(cc)
subplot(2,1,1)
axis([0 1 -150 50])

figure
subplot(2,1,1)
plot(bb)
subplot(2,1,2)
plot(cc)

figure
subplot(2,1,1)
plot(t,'*');
subplot(2,1,2)
plot(bbr,'*');

great post and a very commonly misunderstood subject. There are some numerical issues when flipping zeros of long FIR filters. you can google ‘leja ordering of roots’. this drastically reduces the numerical problems.

 

[j_j]

If you're creating filters tight enough for only one place in time and space, you might as well switch to headphones.

Not sure but if you mean "tight in magnitude accuracy" you shouldn't do that. Exhale and your filters are wrong.

Smoothing of correction filters needs to happen on something like an ERB scale after 300/400 Hz. (No, not schroeder frequency, human hearing issues)

 

[j_j]

great post and a very commonly misunderstood subject. There are some numerical issues when flipping zeros of long FIR filters. you can google ‘leja ordering of roots’. this drastically reduces the numerical problems.

Of course. I was lazy and used short filters and doubles.

 

[DWPress]

Not sure but if you mean "tight in magnitude accuracy" you shouldn't do that. Exhale and your filters are wrong.

I meant filters that are created for the LP only - so yes, exhale or stand up and it goes to hell....

 

[j_j]

I meant filters that are created for the LP only - so yes, exhale or stand up and it goes to hell....

If it's that precise, it will be wrong at both ears if your head has the mic spot in the middle. So yeah. Not a good idea, it will most likely make things worse, actually.

 

[MengW]

There are some information about excess phase and excess group delay in this page Minimum Phase in REW help.
https://www.roomeqwizard.com/help/help_en-GB/html/minimumphase.html.


It is not my major. Do you think it is correct?

For example, Anywhere the excess group delay plot is flat is a minimum phase region of the response.

The excess group delay plot also clearly shows there is a time offset between the subwoofer and the main speaker, the sub being about 25ms delayed, which is not so obvious from the overall group delay plot. Excess group delay is a useful plot for time aligning speakers.

Thanks!

 

[j_j]

There are some information about excess phase and excess group delay in this page Minimum Phase in REW help.
https://www.roomeqwizard.com/help/help_en-GB/html/minimumphase.html.


It is not my major. Do you think it is correct?

For example, Anywhere the excess group delay plot is flat is a minimum phase region of the response.

If the group delay is constant, that's otherwise known as "Linear phase".

If phase is a straight line ( phi = 2 * pi * f *t ), where phi is phase in radians, f Hrz and t in seconds. I did not read the site, I've got a house full of guests and they come first, but if delay is constant, that's "linear phase".

Of course, "minimum phase" means that there are no system zeros (or poles, obviously) outside the unit circle. If there are zeros outside the unit circle, things are different. Room responses are often like that, by the way.

The excess group delay plot also clearly shows there is a time offset between the subwoofer and the main speaker, the sub being about 25ms delayed, which is not so obvious from the overall group delay plot. Excess group delay is a useful plot for time aligning speakers.

Thanks!

 

[bmc0]

For example, Anywhere the excess group delay plot is flat is a minimum phase region of the response.

@j_j doesn't seem to like the term "excess group delay" . Since @JohnPM defines excess group delay as the slope of the difference between the measured phase and the minimum phase (i.e. analogous to normal group delay, but derived from the excess phase), this is true by definition. If the excess group delay is flat within a given interval, the only difference between measured and minimum phase in that interval is a pure time delay component.

Excess group delay is a useful plot for time aligning speakers.

Keep in mind that if you use minimum phase crossover filters, the excess group delay will generally not be flat when things are properly aligned.

 

[j_j]

@j_j doesn't seem to like the term "excess group delay" . Since @JohnPM defines excess group delay as the slope of the difference between the measured phase and the minimum phase (i.e. analogous to normal group delay, but derived from the excess phase), this is true by definition. If the excess group delay is flat within a given interval, the only difference between measured and minimum phase in that interval is a pure time delay component.


Keep in mind that if you use minimum phase crossover filters, the excess group delay will generally not be flat when things are properly aligned.

If it's time delay, just CALL IT TIME DELAY???? FFS.

 

[bmc0]

I think you may have misread what I wrote.