Compensating for mains and sub delay

[dasdoing]

but an all-pass filter might (should?) work in order to only correct the delay. never experimented with this though.
as has been said, the best way is to use a "normal" delay only on the main chanels

 

[dasdoing]

Since you don't have a way of separating subwoofer output in your setup, your only free option is to apply a filter like below in Equalizer APO which will send frequencies below 100Hz around 20ms earlier. Since the filter itself causes a processing delay of 170ms, it MAY not cause pre-ringing. If it does, then you can try delaying above 100Hz by 20ms instead. Let me know if you need any help with that.

View attachment 298446

Here's the group delay graph after the filter is applied to a Dirac pulse:


View attachment 298450

oops, exactly

 

[OCA]

And here's "for certain ringing free" way of doing it by delaying everything after 100Hz:

 

[OCA]

Does AC interference cause group delay? How? If it was a ground loop it would manifest as a 50Hz hum but I don't know how it would cause group delay? I mean, that's a pretty substantial peak on that excess GD plot, and if there was a hum it would surely be audible.

It does (at least in my case) and what is most worrying is it does so randomly!!! L0 and L1 are default 1M length measurements for the same speaker with 5 minute intervals. I get these GD at 50Hz and all its harmonics up to 1kHz

And I should add that I don't have an audible hum!

 

[Keith_W]

Okay, so I have received two interpretations of that peak at 50Hz:

1. It is caused by 50Hz AC (@OCA )
2. It is a non minimum phase region in room response (@dasdoing )

I can understand (2) and I find it plausible. But I still do not understand how a ground loop can increase group delay. If you say that it is intermittent then that would point towards an electrical cause rather than non-minimum phase behaviour of your room (unless your measurements are done with something varying in your room, e.g. open windows or doors? I assume you are keeping things constant). Sorry if I come across as obstinate, it is not my intention. I just don't see how, particularly as you say it is not audible. I would love it if someone knowledgeable provided an explanation.

 

[OCA]

I would love it if someone knowledgeable provided an explanation.

Me, too actually

No change in room conditions during measurements. I have this problem since years in this house. Plugging everything to the same wall socket helped things a bit, at least no longer audible hum but in all frequency domain graphs , I'll see sharp fluctuations around 50Hz and less so in its harmonics.

 

[ernestcarl]

but an all-pass filter might (should?) work in order to only correct the delay. never experimented with this though.
as has been said, the best way is to use a "normal" delay only on the main chanels

Yep. Unfortunately, a lot of people may not have independent time delay controls... say, when using only simple DAC stereo outs -- as many apparently do here in ASR... Phase time correction/inversion appears to work best when one is also able adjust individual speaker/driver time offsets.



I would only caution those starting out to avoid very high Q and deep narrow gain filters.

And, of course, always do before and after listening tests.

all-pass_phase_inverse_test - Google Drive

drive.google.com

 

[dasdoing]

Okay, so I have received two interpretations of that peak at 50Hz:

1. It is caused by 50Hz AC (@OCA )
2. It is a non minimum phase region in room response (@dasdoing )

I can understand (2) and I find it plausible. But I still do not understand how a ground loop can increase group delay. If you say that it is intermittent then that would point towards an electrical cause rather than non-minimum phase behaviour of your room (unless your measurements are done with something varying in your room, e.g. open windows or doors? I assume you are keeping things constant). Sorry if I come across as obstinate, it is not my intention. I just don't see how, particularly as you say it is not audible. I would love it if someone knowledgeable provided an explanation.

3 impulses randomly delayed against each other and than added together

FR and phase

group delay

excess group delay

 

[dasdoing]

should this actually happen? 3 impulses, when the other starts the one in front is allready silent.
so this would be an artifact of the windowing?

 

[dasdoing]

here is a 24dB/o LR 80Hz lowpass delayed 20ms, a coresponding highpass and their sum in the mdtat. https://drive.google.com/file/d/13VxBqsQyoeptRl2U3zFBHezxqzpDribX/view?usp=sharing
in theory the best compensation would be to simply invert the sum. but look at what happens to the FR

here is the resulting excess group delay btw

 

[ErLan]

Just would like to take a sec for an interim thank to all contributers so far! On a glance I see many new ideas and new exploration paths.
I will start diving into correspondence later on, also will provide addional info per request and (most likely) followup questions as well

Best,

 

[ErLan]

Could you show us the measurement?

Attached are typical *.mdat + spectrogram snapshot that triggered my question.
I am only starting, so delay is not the only item to deal with but rather one of the initial ones - hope it answers.

 

[dasdoing]

ok, here is the easiest way to do it with just a few steps in EQ-APO.
assuming this is the exact crossover (if not, easily solved, too)

do same for right channel

 

[OCA]

ok, here is the easiest way to do it with just a few steps in EQ-APO.
assuming this is the exact crossover (if not, easily solved, too)

View attachment 298549

do same for right channel

This is genius but does it actually work? The final copy operation needs to merge L2 with L or else you will end up with a lpf applied L2. I'm asking because I'm not familiar with copy operation in EQ Apo.

 

[dasdoing]

This is genius but does it actually work? The final copy operation needs to merge L2 with L or else you will end up with a lpf applied L2. I'm asking because I'm not familiar with copy operation in EQ Apo.

wait a second.
you are the obsessive compulsive audiophile guy.
I wanted to talk to you anyways soon.
got to know your channel only last week because of a topic in REW forum.
it was your video doing inversions in REW. I was so surprised to find another one doing excessive experiments with it. I will post my aprouch here soon in a seperate topic and mark you.

regarding the copy function. it is litteraly copiing. what is on the copied channel is beeing preserved (unless you define "copy: 0=L L=L2").
I can't see the result on the analysis panel as I made this in a seperate tab (don't want to loose my own config), but I bet the FR is oposite of the sum I did in the Mdat in post 30

 

[dasdoing]

Attached are typical *.mdat + spectrogram snapshot that triggered my question.
I am only starting, so delay is not the only item to deal with but rather one of the initial ones - hope it answers.

thanks, I needed a nice mdat of some other speakers to experiment.
let me tell you something: was creating a phase correction for the left speaker and found a nice window. listened to the combined result over headphones and everything sound correct.
did the same for the right one and I hear pre-delay. I couldn't explain it so I listened to the original IR and it already has pre-delay baked in. how can that be? do you hear any predelay on the right speaker with the intro of Billie Jean for example? anyways, could be an artifact of the measurement. the right way to test the filters is on the real speakers.
I think these should be good: https://drive.google.com/file/d/1Wt57R3kL7Uj_6mDDQjtLM3GUq6Oes_Qn/view?usp=sharing
throw the .wav into the convolver of EQ-Apo. it's 44100

here is left speaker phase before and after

note that this is all experimental. might or not work. give me some feedback please

 

[ErLan]

thanks, I needed a nice mdat of some other speakers to experiment.
let me tell you something: was creating a phase correction for the left speaker and found a nice window. listened to the combined result over headphones and everything sound correct.
did the same for the right one and I hear pre-delay. I couldn't explain it so I listened to the original IR and it already has pre-delay baked in. how can that be? do you hear any predelay on the right speaker with the intro of Billie Jean for example? anyways, could be an artifact of the measurement. the right way to test the filters is on the real speakers.
I think these should be good: https://drive.google.com/file/d/1Wt57R3kL7Uj_6mDDQjtLM3GUq6Oes_Qn/view?usp=sharing
throw the .wav into the convolver of EQ-Apo. it's 44100

here is left speaker phase before and after

View attachment 298625

View attachment 298626


note that this is all experimental. might or not work. give me some feedback please

Thanks!!! I am eager to test it (together with some other ideas brought up here)..
Currently, working to secure some time for a weekend full of swipes

 

[ernestcarl]

Attached are typical *.mdat + spectrogram snapshot that triggered my question.
I am only starting, so delay is not the only item to deal with but rather one of the initial ones - hope it answers.

Condition of your measurements don't look particularly ideal, so this may be the most excess phase correction that's quite "possibly" psychoacoustically benign (needs to be tested):

Also check how the windowing affects the "excess group delay" curves.


Extract the windowed excess phase:

If you want to keep using REW for simplicity (rather than rePhase), invert the excess phase:

Remove the windowing from the original measurement before convolving it with the generalized phase/time "correction" filter:

Before & After:



I would not attempt to pull down anything above 120-140 Hz as that would surely result in much more -- quite possibly "severe" -- audible pre-ringing than what this non-optimal correction filter already does.


-------

*Eh, this particular REW generated filter already does not look all that great post export (all pass/phase inverse filter is just sooo steep in some places):

 

[dasdoing]

Condition of your measurements don't look particularly ideal, so this may be the most excess phase correction that's quite "possibly" psychoacoustically benign (needs to be tested):

View attachment 298676

Also check how the windowing affects the "excess group delay" curves.


Extract the windowed excess phase:

View attachment 298677


If you want to keep using REW for simplicity (rather than rePhase), invert the excess phase:

View attachment 298678


Remove the windowing from the original measurement before convolving it with the generalized phase/time "correction" filter:

View attachment 298679


Before & After:

View attachment 298680 View attachment 298681 View attachment 298682 View attachment 298683

I would not attempt to pull down anything above 120-140 Hz as that would surely result in much more -- quite possibly "severe" -- audible pre-ringing than what this non-optimal correction filter already does.


-------

*Eh, this particular REW generated filter already does not look all that great post export (all pass/phase inverse filter is just sooo steep in some places):

View attachment 298689

View attachment 298687

listen to the IRs, before and after
as I mentioned 2 posts above you can allready hear predelay in the original,
but in your resulting one I can clearly aditionally hear this:

on the kick drum of the Billie Jean intro

 

[ernestcarl]

thanks, I needed a nice mdat of some other speakers to experiment.
let me tell you something: was creating a phase correction for the left speaker and found a nice window. listened to the combined result over headphones and everything sound correct.
did the same for the right one and I hear pre-delay. I couldn't explain it so I listened to the original IR and it already has pre-delay baked in. how can that be? do you hear any predelay on the right speaker with the intro of Billie Jean for example? anyways, could be an artifact of the measurement. the right way to test the filters is on the real speakers.
I think these should be good: https://drive.google.com/file/d/1Wt57R3kL7Uj_6mDDQjtLM3GUq6Oes_Qn/view?usp=sharing
throw the .wav into the convolver of EQ-Apo. it's 44100

here is left speaker phase before and after

View attachment 298625

View attachment 298626


note that this is all experimental. might or not work. give me some feedback please

I have to say, your results look far more remarkable!

Although, I would like to see actual objective verification measurements and subjective listening tests from the OP.