Audibility of group delay at low frequencies

[dasdoing]

oh yes, it is importand to emphasize that "linear phase" correcting software actualy uses mixed phase

 

[hyperplanar]

as always with audio, it's a compromise where you have to find the right point in the middle.
same is for pre-ringing and (to put it into perspective I will call it) pos-ringing.

I agree, but it’s important to note that (especially for bass) preringing is much more audible than postringing, which is a “natural” phenomenon and very well masked. I do have a crossover phase linearization filter in my system for the 2 kHz woofer/tweeter crossover which subjectively provides good results. At such a high frequency it would be extremely difficult to hear preringing for such a smooth filter.

I didn’t have nearly as good of a result linearizing the 80 Hz subwoofer/speaker crossover... The difference was very audible with kick drums and the bass certainly seemed tighter, but at times I felt like I heard the bass of the kicks starting very slightly before the rest of it. But I guess this depends on personal preference and the range of positions you listen to your system in.

personaly when I compare (listening) FIR room correction (actualy mixed phase) filters to filters created with REW in my system there is dozen times bigger impact on center clarity then audible delay. as Bob Katz describes it (he uses Acourate): The diference is like 3D vs 2D. I have seen other describe this as "lost wideness"; this is not the case, the mid signal is more centered (as it was supoused to be), not the side signal

This isn’t because of excess phase correction in the bass region though, right? I’d guess most of this is due to correcting any L/R frequency response imbalances.

 

[dasdoing]

I agree, but it’s important to note that (especially for bass) preringing is much more audible than postringing, which is a “natural” phenomenon and very well masked. I do have a crossover phase linearization filter in my system for the 2 kHz woofer/tweeter crossover which subjectively provides good results. At such a high frequency it would be extremely difficult to hear preringing for such a smooth filter.

I didn’t have nearly as good of a result linearizing the 80 Hz subwoofer/speaker crossover... The difference was very audible with kick drums and the bass certainly seemed tighter, but at times I felt like I heard the bass of the kicks starting very slightly before the rest of it. But I guess this depends on personal preference and the range of positions you listen to your system in.

yea, when you hear it it sounds very bad. "pos-ringing" on the other hand will always be masked by other sounds, plus we are used to it.
I can only hear the improvement in group delay when directly comparing (as you say: tighter kick). we can say that this is something you don't need, but if you invest the time you will have improvement

This isn’t because of excess phase correction in the bass region though, right? I’d guess most of this is due to correcting any L/R frequency response imbalances.

though the bass correction obviously doesn't impact this in the overall image (the bass sure get's centered, too) it is the result of phase correction. there is no big diference in center clarity before and after REW correction, but there is a gigantic diference after FIR correcting; it actualy sounds like if speaker were changed. the image also seams "deeper". the sound seams to have a space instead of a "wall of sound" generated between your speakers.

 

[dasdoing]

btw: Acourate will produce filter processed tracks for trying out their software: https://www.audiovero.de/en/acourate-test-for-free--.php

 

[hyperplanar]

yea, when you hear it it sounds very bad. "pos-ringing" on the other hand will always be masked by other sounds, plus we are used to it.
I can only hear the improvement in group delay when directly comparing (as you say: tighter kick). we can say that this is something you don't need, but if you invest the time you will have improvement




though the bass correction obviously doesn't impact this in the overall image (the bass sure get's centered, too) it is the result of phase correction. there is no big diference in center clarity before and after REW correction, but there is a gigantic diference after FIR correcting; it actualy sounds like if speaker were changed. the image also seams "deeper". the sound seams to have a space instead of a "wall of sound" generated between your speakers.

Yeah, I also found my speakers to subjectively sound “deeper” with the woofer/tweeter crossover phase linearization.

But with regards to your comment about the center becoming sharper... did your REW and FIR filters have the exact same frequency response or were they different?

 

[dasdoing]

Yeah, I also found my speakers to subjectively sound “deeper” with the woofer/tweeter crossover phase linearization.

But with regards to your comment about the center becoming sharper... did your REW and FIR filters have the exact same frequency response or were they different?

they are indeed different. it is not possible to match the correction with REW with it's limitation to 20 eq filters. I could actualy compare this better with my software (https://www.ohl.to/about-audio/audio-softwares/align2) since it can generate minimum phase correction too. got me curious, will do this after the gym

 

[bigjacko]

From you guys talk correcting phase can only correct on axis but not off axis? Why is that possible if phase is just timing?

 

[andreasmaaan]

From you guys talk correcting phase can only correct on axis but not off axis? Why is that possible if phase is just timing?

It depends which axis. Assuming drivers are vertically arranged, phase correction can result in linear phase across the entire horizontal plane*, but as one moves up or down, causing the relative distances between the listener and each driver to change, timing differences will result from path length differences, and phase will of course no longer be linear.

*Assuming of course drivers are also aligned in the Z-plane, which is arguably rarely the case, but let's not overcomplicate it here.

 

[dasdoing]

they are indeed different. it is not possible to match the correction with REW with it's limitation to 20 eq filters. I could actualy compare this better with my software (https://www.ohl.to/about-audio/audio-softwares/align2) since it can generate minimum phase correction too. got me curious, will do this after the gym

So I just compared.

ignoring a small variation above 14Khz the FR responses match
(var-smoothing > right channel above, left below (there are 4 FRs > left and right offset for visualization))

they do sound TOTALY diferent though in terms of center image. the MP corrected one is much more spreaded.

this is the right channel after MP correction. in black the exess phase

this is the right channel after LP correction. if you ignore the vertical jumps caused by dips it is close to linear phase (I am sure I can still improve this. I aplied delay to the mains, which I should "delegate" to the FIR correction, too)

From you guys talk correcting phase can only correct on axis but not off axis? Why is that possible if phase is just timing?

as soons as you move the mic/head the relation of delay of direct and reflected sound will change.
following image is an example of a speaker but it explains it in the room, too.

 

[andreasmaaan]

So I just compared.

ignoring a small variation above 14Khz the FR responses match
(var-smoothing > right channel above, left below)

View attachment 98273

they do sound TOTALY diferent though in terms of center image. the MP corrected one is much more spreaded.

this is the right channel after MP correction. in black the exess phase

View attachment 98274

this is the right channel after LP correction. if you ignore the vertical jumps caused by dips it is close to linear phase (I am sure I can still improve this. I aplied delay to the mains, which I should "delegate" to the FIR correction, too)

View attachment 98275





as soons as you move the mic/head the relation of delay of direct and reflected sound will change.
following image is an example of a speaker but it explains it in the room, too.


View attachment 98276

Could you please put the FR on a 50dB vertical scale pls?

 

[dasdoing]

Could you please put the FR on a 50dB vertical scale pls?

like this? the diference is very minimal; ignoring above 14khz which shouldn make a diference

 

[dasdoing]

I could actualy try to make this diference audible to you guys on a headphone by convolving music with my room IRs and the filters. not sure how this will sound on headphones though

 

[andreasmaaan]

I could actualy try to make this diference audible to you guys on a headphone by convolving music with my room IRs and the filters. not sure how this will sound on headphones though

That would be interesting, thanks.

like this? the diference is very minimal; ignoring above 14khz which shouldn make a diference

I guess it depends how you define minimal The differences in amplitude there look greater than established audibility thresholds (would be clearer if you lined the graphs up directly over each other).

The phase graphs look pretty weird to me, TBH. Would you mind trying two things?

1. Put some smoothing in place so that all the narrow peaks and dips in the FR disappear.
2. Unwrap the phase.

 

[dasdoing]

That would be interesting, thanks.



I guess it depends how you define minimal The differences in amplitude there look greater than established audibility thresholds (would be clearer if you lined the graphs up directly over each other).

The phase graphs look pretty weird to me, TBH. Would you mind trying two things?

1. Put some smoothing in place so that all the narrow peaks and dips in the FR disappear.
2. Unwrap the phase.

I give you the mdat https://drive.google.com/file/d/1QIQIT2-t22IyxJkVKgwPOl1IlYxm5o-T/view?usp=sharing

if you send me IRs of your room I could actualy create a filter for you tomorrow to test it out. though I am not sure if my settingas will work in your room since mine is treated

 

[andreasmaaan]

I give you the mdat https://drive.google.com/file/d/1QIQIT2-t22IyxJkVKgwPOl1IlYxm5o-T/view?usp=sharing

Thanks

EDIT: post deleted as it was based on idiocy.

 

[andreasmaaan]

Oh, one thing though. You were 100% right that the amplitude responses were very similar:

 

[vavan]

You were 100% right that the amplitude responses were very similar

perhaps using trace arithmetic (a/b) to show the diff would be even more evident

 

[ernestcarl]

They both look very similar to me in terms of group delay TBH.

I don't think the fourier graph is best for what we're trying to see here... but I agree.

Group Delay view

Spectrogram wavelet view

There is an improvement, yes, but not that big -- I'm skeptical of how much of a practical audible improvement this makes.

There's some weirdness in the high frequencies decay which can be seen in the GD, decay, and fourier views for both minimum phase and linear phase responses.

Clarity traces are slightly improved with linear phase, but again, not that much of a difference.

Eh... regardless of minimum or linear phase type corrections being used, the steady-state still looks over-corrected -- even in a nearfield setup situation. But that is just what I think personally. I claim no expertise whatsoever.

 

[andreasmaaan]

There's some weirdness in the high frequencies decay which can be seen in the GD, decay, and fourier views for both minimum phase and linear phase responses.

Clarity traces are slightly improved with linear phase, but again, not that much of a difference.

That looks very different to mine! Now I'm wondering if I've used incorrect settings in REW (not a piece of software I normally use). Which program have you used there to generate those graphs?

 

[ernestcarl]

That looks very different to mine! Now I'm wondering if I've used incorrect settings in REW (not a piece of software I normally use). Which program have you used there to generate those graphs?

In the spectrogram tab

You can change the appearance settings as well -- just change it to whatever method or resolution is needed to see things clearer.

FDW doesn't affect the spectrogram at all, but it can clear up some of the reflection-induced rotations in seen in the phase traces.

From REW's help doc:

Fourier or Wavelet.

In Fourier mode the plot uses fixed width windows, which mean the plot has the same time resolution at all frequencies. If the plot spans a wide range of frequencies this usually means the time resolution is either too low at high frequencies or too high at low frequencies. A 100 ms window, for example, gives 10 Hz frequency resolution. At low frequencies that is a big octave fraction (1/1.4 octaves at 20 Hz), at high frequencies a very, very small octave fraction (1/1386 octaves at 20 kHz).

For a time-frequency plot it would be more useful if the tradeoff between time and frequency resolution varied with frequency, using a constant octave fraction for frequency resolution rather than a constant number of Hz and so giving higher time resolution at high frequencies and lower at low frequencies. A wavelet transform can achieve that, specifically a constant Q Continuous Wavelet Transform (CWT). A constant Q wavelet transform is mathematically equivalent to using a frequency-dependent window to produce the spectrogram, which is what REW does.


---

The wavelet view is a favorite of mine. You can see reflections and resonances found in the impulse and ETC, and the broken solid track lines also seem to correlate well with the actual phase itself (best when zoomed-in really close).