PEQ and a sound stage

[Alp]

Hey there fellow room eqers!

So I got RME ADI-2 with 5 to 7 PEQ filters, bought an Umik and spend about 2 days crafting the best charts and understanding speaker placement. I have a full size speakers and great stereo system with ok room treatment (rather lacking).

I got a lot of nice charts that follows my target curve. I even tried double eq - separate PEQ for left and right speaker.

The problem: soundstage suffered. With single eq for both speakers stage collapsed in one point with some minor imaging. With separate eqs for both speakers it is better, but worse than no eq.

I don't understand how it works. Even if I try to keep main chart peeks and valleys similar, I can't get it to sound like with no eq.
I mainly played in the 130-1000 range to get the nice flat response there
This leaves PEQ as a simple bass/treble adjust and pretty much a toy otherwise. What's your experience? I can share measurements if someone is interested to debug the sound stage challenge.

 

[radix]

In the 130-1000 range, some of that might be room modes and some of that might be reflections. Could you post your measurements with a flat EQ and after EQ?

Sometimes, soundstage is an illusion caused by the speaker reflection off the L and R walls, which makes it sounds like its a wider space than it really is. The EQ might have toned down those frequencies so you lost that illusion. I'm just guessing here.

You can also look at the waterfall or RT60 charts to see how much off-path or reverberant components you have.

 

[Alp]

Here it is. Measures of L and R as well as L+R before and after eq included eq filters as well.

Note I am aware of L speaker placement issues, unfortunately can't move it much

 

[radix]

I'm not clear on what's what. It looks like the "part2" file has some "no eq" traces. Is that your baseline? And then when you apply some filter (which one?) the soundstage suffers?

Here's some general comments about EQ. I think you know some of this, but I just want to put it out there to be clear.

For hifi playback, one generally wants a target like the Harman curve. You can find curves here for REW. The curve can be adjusted based on preference (I usually like more bass, for example).

A collection of speaker target responses in .csv/.txt format

In order to use REW's automatic EQ feature, I started collecting and converting every speaker target that I could find into the necessary .txt format. In this thread I want to share my work so far, as well as ask you guys for target responses that I may have missed. Comparison: So...

www.audiosciencereview.com

From your PEQs, it looks like you were going for a Harman?

I see in your PEQ filters you have a fair bit of boost. Maybe +5 or +8 dB depending on L or R. Generally you don't want to boost much. You cannot usually dig yourself out of a hole, you just get a bigger hole. These holes are usually caused by interfering waves that cancel each other. If you make them louder, you just make more cancellation. Some dips, e.g. at the speaker crossover, you can boost

When you boost, you need to make sure the DSP has enough headroom to not clip. I think the RME has a lot of internal headroom, but I'm not sure it will like an 8dB boost.

Usually you only want to cut, or maybe boost a tiny bit. I usually get decent suggestions from the REW EQ tool loading a target curve, fitting the response to the target, then optimizing the target.


For example, here's what I got for your "L no eq 27." Notice that the bass is still pretty choppy. I'd try to find out what's causing all those nulls and fix it. You're not going to be able to EQ out of those.

This is just an example, and not meant to be an RME PEQ setting. I usually generate those in VirtuixCAD where I can setup the exact 5+2 PEQ of the RME.

Filter Settings file Room EQ V5.30.8 Dated: Mar 30, 2024 10:58:44 PM Notes: Equaliser: Configurable PEQ L no eq 27 Filter 1: ON PK Fc 27.49 Hz Gain -4.20 dB Q 13.49 Filter 2: ON PK Fc 50.00 Hz Gain -3.20 dB Q 7.97 Filter 3: ON PK Fc 83.98 Hz Gain -10.10 dB Q 10.63 Filter 4: ON PK Fc 159.4 Hz Gain -4.60 dB Q 24.97 Filter 5: ON PK Fc 241.1 Hz Gain -8.00 dB Q 3.40 Filter 6: ON PK Fc 467.5 Hz Gain -7.50 dB Q 3.29 Filter 7: ON PK Fc 662.0 Hz Gain -6.70 dB Q 7.86 Filter 8: ON PK Fc 789.0 Hz Gain -7.80 dB Q 6.49 Filter 9: ON PK Fc 2387 Hz Gain -4.50 dB Q 1.06 Filter 10: ON PK Fc 7226 Hz Gain -3.20 dB Q 3.59

An EQ like this is going to take away from the lower mid and presence region. It will sound different. It will also be a fair bit quieter and you might need to boost the volume some to compare.

I set the filter target to 75dB, which is maybe a little low. Could have been maybe 78, not as flat, but less cut.

I'm not sure if that helped.

If you can find what is causing the 1.5 kHz dip and fix that with room treatments, you could then avoid a lot of cutting. It might be a 1st reflection.

Marc

 

[Alp]

Thank you Marc. I indeed used Harman curve. "No eq" measures are reference measurements without any eq. "eq" measurements are with "optimized" filters applied. "R fix eq" and "L+R fix eq" has R mirrored eq applied.

In this sample, every channel was eqed separately: L with L optimized and R with either R optimized or R mirrored

As you can see from "L eq" or "R eq" or "L+R eq" the boost worked fine and eliminated the gaps.

I used 76db baseline. Will try later today with absolutely no boosts

 

[Pareto Pragmatic]

I used 76db baseline. Will try later today with absolutely no boosts

That will help.

Whatever you end up with, if you still have the problem, then I suggest turning all your PEQ filters off, then turning them back on one at a time starting with the lowest frequency and then adding the rest one by one. Try to see if it is one particular setting, or if it starts to crop up at some specific frequency and gets worse as you add more on top of that one.

 

[Alp]

@radix regards the eqs you provided. RME has limited controls. 1 and 7 PEQs are bass and treble with min q 0.5 and max 1.5. they are also not PK filters, but LS and HS respectively (as I understand). Filters 1 to 3 have max q of 9.9. 4 to 5 have max Q of 5. Max boost/cut is 12 db. It has frequency precision of 1Hz until 200Hz, then it is 10Hz, and after 2kHz or so a step is 100Hz

You said that you setup RME filters in that application, but from your post I see unrelated set of eqs that can't be applied on adi-2 DAC devices

 

[Alp]

Separate EQ for L and R channels destroys stereo image: confirmed

Method: used EQ boost and EQ cut to make response curve flat separately for L and R. Confirmed with follow up measurements of EQ applied via ADI-2 DAC.
Used test Audiofile specifically designed to test stereo image.
Locked autoref on RME, pushed volume to -6 dbr to compensate for EQ boost.

Observation: without EQ image is sharp, precise and centered. With EQ image is muddy and shifted to right and top.

Next tests: same EQ for left and right, EQ only in bass area, same tests, but limit boost/cut under 5db and q under 3

Unfortunately, PEQ in RME implementation is destructive to be used as room correction in mid-high area

 

[czt]

Minimum-phase filters cause phase rotation (no way out, linear-phase filters cause pre-ringing). EQ is viable, if it corrects larger problems than the side effects caused by it. (Something very weird going on with your bass range.)

 

[rogier]

The problem with EQ-ing both speakers differently, I believe, is that PEQ filters usually change not only the magnitude dependent on the frequency, but also the phase. Unless you’re very careful, if you use different filters for left and right, the two speakers will be out of phase in some frequency ranges.

 

[LIΟN]

How far are you listening?
recommend gating any EQ above roughly 200hz rather than full decay (REW defaults to 500ms), and judging whether to apply it with caution.
Let's apply 1/6 smoothing for ease of viewing.

It's your 500ms FR.

and this is 5ms FR.

and this is 1ms FR.
I don't think you need a lot of EQ on the highs, it would help to move or clean up anything that creates a response difference of 1-2ms next to the speaker.
Also, a target curve is both a goal and an outcome.
The target curve is naturally created by the position and distance of your speakers and the size of your room.
So you can listen to it yourself (roughly above 500hz) and apply some shelf filter and PEQ to adjust it as much as you like, but I don't recommend EQing based on the graph alone.

Rather, I think it's good that you focus on the bass area.
As you increase the time of the gating, the response of the bass will seem to rise due to the influence of the room, but it won't look as clean.

You may find the imbalance in Decay more audibly unpleasant than the visible response.
I don't know what system you are running, but it would be nice to manage your room modes with at least two subwoofers.

So going back to the original PEQ story, I think it's better not to use EQs with fine-grained Q values.
I would suggest just a smooth balance between 50 and 300Hz with a maximum of Q value 3.

Otherwise, For non-symmetric cases, you can also try using a DRC like DiracLive (or create your own FIR filter) to improve imaging.
Dirac is not magic that makes everything perfect, but the problem with asymmetrical rooms (or settings) is that they sound very strange with a left-right mismatch between direct and early reflections. Sometimes it's skewed to the left, sometimes to the right.
It can be somewhat compensated for, so you might want to consider it.

 

[Purité Audio]

Separate EQ for L and R channels destroys stereo image: confirmed

Method: used EQ boost and EQ cut to make response curve flat separately for L and R. Confirmed with follow up measurements of EQ applied via ADI-2 DAC.
Used test Audiofile specifically designed to test stereo image.
Locked autoref on RME, pushed volume to -6 dbr to compensate for EQ boost.

Observation: without EQ image is sharp, precise and centered. With EQ image is muddy and shifted to right and top.

Next tests: same EQ for left and right, EQ only in bass area, same tests, but limit boost/cut under 5db and q under 3

Unfortunately, PEQ in RME implementation is destructive to be used as room correction in mid-high area

You shouldn’t be adding any filters in the non -minimal phase region of the FR ie the mids and highs.
REW can you show you the minimal phase regions which are generally in the low bass.
Keith

 

[Alp]

Bass eq under 70 Hz seems to not affect stereo image most likely due to how we have issues identifying bass sources anyway

 

[Alp]

How far are you listening?
recommend gating any EQ above roughly 200hz rather than full decay (REW defaults to 500ms), and judging whether to apply it with caution.
Let's apply 1/6 smoothing for ease of viewing.

View attachment 360350

It's your 500ms FR.

View attachment 360352

and this is 5ms FR.

View attachment 360353

and this is 1ms FR.
I don't think you need a lot of EQ on the highs, it would help to move or clean up anything that creates a response difference of 1-2ms next to the speaker.
Also, a target curve is both a goal and an outcome.
The target curve is naturally created by the position and distance of your speakers and the size of your room.
So you can listen to it yourself (roughly above 500hz) and apply some shelf filter and PEQ to adjust it as much as you like, but I don't recommend EQing based on the graph alone.

View attachment 360355

Rather, I think it's good that you focus on the bass area.
As you increase the time of the gating, the response of the bass will seem to rise due to the influence of the room, but it won't look as clean.

View attachment 360357

You may find the imbalance in Decay more audibly unpleasant than the visible response.
I don't know what system you are running, but it would be nice to manage your room modes with at least two subwoofers.

So going back to the original PEQ story, I think it's better not to use EQs with fine-grained Q values.
I would suggest just a smooth balance between 50 and 300Hz with a maximum of Q value 3.

Otherwise, For non-symmetric cases, you can also try using a DRC like DiracLive (or create your own FIR filter) to improve imaging.
Dirac is not magic that makes everything perfect, but the problem with asymmetrical rooms (or settings) is that they sound very strange with a left-right mismatch between direct and early reflections. Sometimes it's skewed to the left, sometimes to the right.
It can be somewhat compensated for, so you might want to consider it.

Thank you a lot for the analysis!

I am listening from 2.7m from each speaker.

I figured out that I don't hear much of issues with EQ under 70. So will focus on cutting some 27Hz, 67(72)Hz. Will try the low Q in mids as well

 

[Sokel]

Bass eq under 70 Hz seems to not affect stereo image most likely due to how we have issues identifying bass sources anyway

Use this as a guide:

What is your favorite house curve

OK. Next question for you folks. As far as I can tell there is no spinorama for the Focal Aria 926's, which happen to be my main speakers. However, Amir did do a spinorama for the Aria 906 standmount speakers. They use the same tweeter and mid driver as the 926, but lack the two woofers as they...

www.audiosciencereview.com

 

[er|κzvio1in]

This is just an example, and not meant to be an RME PEQ setting. I usually generate those in VirtuixCAD where I can setup the exact 5+2 PEQ of the RME.

View attachment 360312

Looking at this picture, my idea would be to turn up the bass control on the amplifier (a lot), then shave the peaks with high Q's below 100 hz per speaker. Then some low Q PEQs on all speakers ironing out the region between 200-800. Then I wouldn't touch the rest and use any target curve you prefer. Simple but effective. I'll gladly take any critiques because I'm still learning.

 

[radix]

@radix regards the eqs you provided. RME has limited controls. 1 and 7 PEQs are bass and treble with min q 0.5 and max 1.5. they are also not PK filters, but LS and HS respectively (as I understand). Filters 1 to 3 have max q of 9.9. 4 to 5 have max Q of 5. Max boost/cut is 12 db. It has frequency precision of 1Hz until 200Hz, then it is 10Hz, and after 2kHz or so a step is 100Hz

You said that you setup RME filters in that application, but from your post I see unrelated set of eqs that can't be applied on adi-2 DAC devices

Yes, as I said, this was just an example.

I use VituixCAD for RME filters. This thread give several examples of this approach.

Designing headphone EQ with VituixCAD

I'm not active on this forum, so I apologize if this is old news, but I wanted to share an easy method to provide some EQ insight using the measurement data from ASR in VituixCAD. I started down this path having a few pairs of headphones that have been reviewed here or at least very similar to...

www.audiosciencereview.com

You can set up the optimization for each possible 5+2 filter. This is an example for some headphones.

 

[Solveit]

Separate EQ for L and R channels destroys stereo image: confirmed

No!
Separate EQ for L+R and gain and EQ match them increases the stereo image.

 

[Alp]

No!
Separate EQ for L+R and gain and EQ match them increases the stereo image.

As I mentioned before I don't understand how it works, people in this thread already suggested that PEQ may reverse the phase and do other damage to the signal.
I heard it myself and that is enough for me: I don't see anything that should damage the stereo image on the measurements I provided, but blind testing the results proofs that in my room eq in mid-high area with high Q is destroying stereo imaging.

If you can, could you please eq them to improve the image, I will gladly test it!

I am new to EQ so I may not understand a lot of things

 

[Alp]

I am grateful for all your suggestions people, please keep them coming! Also reading the references shared in this thread, quite a few things to digest!