• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

rePhase output question

USER

Addicted to Fun and Learning
Forum Donor
Joined
Mar 30, 2019
Messages
886
Likes
1,409
So I am trying to incorporate FIR filters to my DSP set-up. I am wondering if these results seem right. The windowed impulse measurement at the end seems off to me and I am wondering if there is an issue. Speakers are Revel M16s.

These are the broad measurements for comparison between the original, basic eq, and, finally, my attempt at using rePhase.

Note that I am averaging several measurements and so the working "average" and "vector average" frequency responses (used for basic eq and my rePhase eq, respectively) are different, hence slightly different results in final FR.

ORIGINAL LEFT.jpg

BASIC EQ LEFT.jpg
REPHASE LEFT.jpg



However when I look closely at the impulse I see this: (original, basic eq, and rePhase):

ORIGINAL IMPULSE LEFT.jpg

BASIC EQ IMPULSE LEFT.jpg

REPHASE IMPULSE LEFT.jpg


is the last image with rePhase OK with some sort of noise reaching the mic before the signal?

Finally, this is what I am working with in rePhase.
REPHASE SETTINGS.jpg


AND ORIGINAL.
REPHASE NON MOD.jpg



THANKS!!!
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
So I am trying to incorporate FIR filters to my DSP set-up. I am wondering if these results seem right. The windowed impulse measurement at the end seems off to me and I am wondering if there is an issue. Speakers are Revel M16s.

These are the broad measurements for comparison between the original, basic eq, and, finally, my attempt at using rePhase.

Note that I am averaging several measurements and so the working "average" and "vector average" frequency responses (used for basic eq and my rePhase eq, respectively) are different, hence slightly different results in final FR.

View attachment 150501
View attachment 150502View attachment 150503


However when I look closely at the impulse I see this: (original, basic eq, and rePhase):

View attachment 150505
View attachment 150506
View attachment 150507

is the last image with rePhase OK with some sort of noise reaching the mic before the signal?

Finally, this is what I am working with in rePhase.
View attachment 150509

AND ORIGINAL.
View attachment 150510


THANKS!!!


Is there any reason why you had to use such high Q phase filters? I usually use 2 as my max. It may result in ringing — whether it’s audible or not is up for your ears to decide. Smoothing the phase response via frequency dependent windowing is what I would do first to avoid “micro-correcting” stuff that’s best left alone. You can go back and use a non-windowed response for the magnitude corrections anytime you wish.

With the graphic EQ, “proportional” is the one compatible with REW’s generic equalizer and not “constant”.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
The ff. is just a graphic example of what I've done for my right surround channel:

1630380200878.jpeg

*please just ignore the somewhat non-sensical acoustic foam absorbers scattered all over the room


Single measurement (uncorrected) in the center MLP
1630380339748.png



Frequency Dependent Windowing applied:

1630382144330.png


1630380485099.png



According the specs, xo is at 1725 Hz so I applied a single filter
1630380396594.png



Single measurement at MLP
1630380513500.png

*You can immediately stop at this stage with the "phase correction"


For the "finer" correction filters, I vector averaged 23 IRs across the entire width of the couch for this single speaker channel (less is obviously possible, but this is mainly to satisfy my OCD).


1630380679236.png


16 additional paragraphic phase EQs are used -- none of which go beyond a Q of 2.


Single measurement at MLP (FDW 15 cycles)
1630380756341.png



Zoomed in impulse responses:

1630380790400.png


After-effect should only be a small precausal bump/dip visible prior to the peak.


However, we can see a much bigger change in the step response (primarily of the midwoofer)
1630381604226.png



I would listen to some test kick-drum and/or snare drum samples with and without the convolution filter to determine whether any significant audible degradation can be heard -- soloing the channel as well as summed with the sub(s) for good measure.


*You can correct just for the excess phase to keep the step response looking a bit more intact afterwards and less mangled -- export "excess phase version" is found in the ALL SPL Measurement Actions window, but in this example I did not.
 
Last edited:
OP
USER

USER

Addicted to Fun and Learning
Forum Donor
Joined
Mar 30, 2019
Messages
886
Likes
1,409
So I think I made some progress today. Thank you all for the tips.
BEFORE LEFT CHANNEL FR AND PHASE
BEFORE FR PHASE.jpg

AFTER:
AFTER FR PHASE.jpg

BEFORE LEFT IMPULSE
BEFORE IMPULSE.jpg

AFTER
AFTER IMPULSE.jpg


I DO THINK I CAN WORK ON MY RIGHT CHANNEL A BIT MORE (IMPULSE SEEMS FINE THO):
BEFORE
R BEFORE FR PHASE.jpg

R AFTER FR PHASE.jpg

R AFTER IMPULSE.jpg
 
Last edited:

Ron Texas

Master Contributor
Joined
Jun 10, 2018
Messages
6,078
Likes
8,916
OP, I thought that I had a nasty room. A 7 db bump at 150 hz is a challenge.
 
OP
USER

USER

Addicted to Fun and Learning
Forum Donor
Joined
Mar 30, 2019
Messages
886
Likes
1,409
OP, I thought that I had a nasty room. A 7 db bump at 150 hz is a challenge.

Oh yeah it is. The set-up is a pair of Revel M16s, one at each side of an unused fireplace, with some chairs by their back corners. It's the best I can do in this room.

The only thing that I have found to take care of those kinds of dips between 100-250 are front-ported floorstanders.

The Revels do sound pretty darn good nonetheless!
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
So I think I made some progress today. Thank you all for the tips.
BEFORE LEFT CHANNEL FR AND PHASE
View attachment 150778
AFTER:
View attachment 150779
BEFORE LEFT IMPULSE
View attachment 150784
AFTER
View attachment 150781

I DO THINK I CAN WORK ON MY RIGHT CHANNEL A BIT MORE (IMPULSE SEEMS FINE THO):
BEFORE
View attachment 150782
View attachment 150783
View attachment 150789

The high frequencies in the right channel starts too early -- i.e. multiple phase rotations going the other way around past 2 kHz.

Either you are using the wrong channel output for the acoustic time reference, or you need to enable the clock time adj. or loopback delay reference within REW's options analysis tab.

Besides that, your graphs already are looking much more "correct" now. ;)
 
Top Bottom