Is DIRAC using the right thing in a 2.2 set up?

[Geertidow]

The 70 Hz frequency often causes a boomy sound in subwoofers, which is why I emphasize addressing this issue to avoid excessive bass resonance.

Here are two test tracks you can use to identify and reduce that boomy 70 Hz frequency. These tracks feature the ~70 Hz range at a relatively high amplitude. Boosting the bass without caution may amplify this frequency, resulting in an overly boomy and unpleasant sound:

To set up your subwoofers properly, first turn them off and listen to these tracks through your R3 speakers to establish a reference of the low frequencies (say, 40 - 100 Hz) that can or will be affected by the subwoofers, and to understand how these tracks should sound. Alternatively, use high-quality headphones for this purpose. Then, turn on the subwoofers, apply your settings, and ensure they add clean, quality bass that complements your speakers without introducing boomy or annoying low frequencies.
After comparing the sound with and without the subwoofers, you may notice that with the subwoofers the sound actually becomes less satisfying. If so, troubleshoot and address any issues to optimize the sound quality to your liking.

Thanks I have made

On the left side, both your speaker and subwoofer exhibit a null at 69 Hz. However, this shouldn't be a major concern, as the right side will partially compensate for it.

Here’s my suggested approach. It’s not necessarily the definitive solution, nor does it guarantee you’ll prefer it, but it’s what I would try first.

I wouldn’t apply EQ to the R3s (except setting crossover frequency on them), as their frequency response is already quite good. Instead, I’d focus on EQing the subwoofers only.

You can keep the crossover frequency at 66 Hz for both the subwoofers and speakers. Personally, I prefer a slightly lower crossover frequency, around 58–60 Hz, since my speakers handle frequencies above 30 Hz well. The rationale is that many subwoofer enthusiasts, myself included, often boost the bass. If the crossover frequency is set at 70 Hz or higher (or even slightly below), this boost can affect the 70–80 Hz range, which often sounds "boomy." A lower crossover helps avoid this issue.

Additionally, I recommend using a steep filter curve, such as a Butterworth 48 dB/octave. This ensures that the subwoofers and speakers only reproduce their intended frequency ranges. A shallow filter curve allows some unwanted frequencies to bleed through. For example, setting the crossover to 58–60 Hz with a 48 dB/octave curve effectively suppresses frequencies around 70 Hz and above.

Here’s how your crossover settings should look:

View attachment 478217
But again, personally, I would set the crossover frequency lower, to 58–60 Hz.

Next, I opened your "Total system, no Dirac" measurement in REW and accessed the EQ window. I set the target level to 90 dB, which is significantly higher than the estimated or average level. I assumed you might want to boost the bass, as many enthusiasts, including myself, enjoy this enhancement.

I then generated the filters using the "Match response to target" option and saved them to a file ("Save filter coefficients to a file", selecting miniDSP and 48 kHz in the popup window).

Here’s the generated filter file: http://ug.link/pashka/filemgr/share-download/?id=ce3d17c2130a4fcebc586d1a37fb2162

In miniDSP, select "PEQ" for your first subwoofer and lick the "Menu" button, then link your first sub to the second sub. Then, choose "Load Biquads File," and upload the generated filter file linked above.
You should see the EQ graph change according to the filters.
View attachment 478218
You might be satisfied with the results as they are. If you want to increase the bass further, simply adjust the gain on your subwoofers, like I often do:
View attachment 478219

Let me know if you like the results.

I have implemented this with a sharp crossover at 60 hz, do you think the mains (plugged ports) can handle a 60 hz 48 db/oct crossover?
I will start listening to the test tracks.

 

[Geertidow]

On the left side, both your speaker and subwoofer exhibit a null at 69 Hz. However, this shouldn't be a major concern, as the right side will partially compensate for it.

Here’s my suggested approach. It’s not necessarily the definitive solution, nor does it guarantee you’ll prefer it, but it’s what I would try first.

I wouldn’t apply EQ to the R3s (except setting crossover frequency on them), as their frequency response is already quite good. Instead, I’d focus on EQing the subwoofers only.

You can keep the crossover frequency at 66 Hz for both the subwoofers and speakers. Personally, I prefer a slightly lower crossover frequency, around 58–60 Hz, since my speakers handle frequencies above 30 Hz well. The rationale is that many subwoofer enthusiasts, myself included, often boost the bass. If the crossover frequency is set at 70 Hz or higher (or even slightly below), this boost can affect the 70–80 Hz range, which often sounds "boomy." A lower crossover helps avoid this issue.

Additionally, I recommend using a steep filter curve, such as a Butterworth 48 dB/octave. This ensures that the subwoofers and speakers only reproduce their intended frequency ranges. A shallow filter curve allows some unwanted frequencies to bleed through. For example, setting the crossover to 58–60 Hz with a 48 dB/octave curve effectively suppresses frequencies around 70 Hz and above.

Here’s how your crossover settings should look:

View attachment 478217
But again, personally, I would set the crossover frequency lower, to 58–60 Hz.

Next, I opened your "Total system, no Dirac" measurement in REW and accessed the EQ window. I set the target level to 90 dB, which is significantly higher than the estimated or average level. I assumed you might want to boost the bass, as many enthusiasts, including myself, enjoy this enhancement.

I then generated the filters using the "Match response to target" option and saved them to a file ("Save filter coefficients to a file", selecting miniDSP and 48 kHz in the popup window).

Here’s the generated filter file: http://ug.link/pashka/filemgr/share-download/?id=ce3d17c2130a4fcebc586d1a37fb2162

In miniDSP, select "PEQ" for your first subwoofer and lick the "Menu" button, then link your first sub to the second sub. Then, choose "Load Biquads File," and upload the generated filter file linked above.
You should see the EQ graph change according to the filters.
View attachment 478218
You might be satisfied with the results as they are. If you want to increase the bass further, simply adjust the gain on your subwoofers, like I often do:
View attachment 478219

Let me know if you like the results.

Interesting to use a sharp roloff, often the recommendation for KEF R3's is to use a low roloff (12 db/oct) around 60 Hz to use the natural roloff of the speakers themselves.
I have implemented the filters and are listening to them now (not the test tracks, first the music I know well).
Also interesting to PEQ the subwoofers together. Given all the options of the MiniDSP flex, wouldn't it be more logical to make seperate filters for each sub (or does it not matter all that much).
Thanks and I'll get back on the results.

Any thoughts on how well DIRAC could be able to do this job?
I do like the idea that it takes multiple measurements in to account.

 

[ViperDom]

I use a target curve with +6dB rolloff at the lower notes.

I call that boost or shelf, not a rolloff.

Does regulating the bass using DSP always mean losing/diminishing the very low notes?

No.

Additionally, I recommend using a steep filter curve, such as a Butterworth 48 dB/octave. This ensures that the subwoofers and speakers only reproduce their intended frequency ranges. A shallow filter curve allows some unwanted frequencies to bleed through.

+1 on the Steep 48X

Interesting to use a sharp roloff, often the recommendation for KEF R3's is to use a low roloff (12 db/oct) around 60 Hz to use the natural roloff of the speakers themselves.

I'm pretty sure that general info on this (that is copy/pasted in speaker manuals and forum info) is mostly applicable to common & older NON-DSP setups which are limited by attempting to "match" a subwoofers fixed crossover.
When you have a MiniDSP it breaks these rules to your advantage.

Verify you are using the LFE input on those SVS's to bypass their internal crossover and let MiniDSP do its thing.

Nice setup BTW. I'm a big fan of the 2.2 KEF setup.

 

[PashkaTLT]

Verify you are using the LFE input on those SVS's to bypass their internal crossover and let MiniDSP do its thing.

Yes, if you use a miniDSP crossover, then make sure the crossover on the subs is set to max/off.

 

[PashkaTLT]

I have implemented this with a sharp crossover at 60 hz, do you think the mains (plugged ports) can handle a 60 hz 48 db/oct crossover?
I will start listening to the test tracks.

I may have missed it, but why did you plug the ports on your speakers?

Your speakers should be capable of handling 60 Hz and even less (depending on how much difference from flat response we allow, in db), though this is near their distortion-free limit. I recommend experimenting with different crossover frequencies to find what sounds best to you.

However, if you set the subwoofer to play frequencies above 70 Hz and then boost its output, this will also amplify the 70 Hz range. You can address this by applying additional filters to the subwoofer using miniDSP.

 

[PashkaTLT]

Interesting to use a sharp roloff, often the recommendation for KEF R3's is to use a low roloff (12 db/oct) around 60 Hz to use the natural roloff of the speakers themselves.
I have implemented the filters and are listening to them now (not the test tracks, first the music I know well).
Also interesting to PEQ the subwoofers together. Given all the options of the MiniDSP flex, wouldn't it be more logical to make seperate filters for each sub (or does it not matter all that much).
Thanks and I'll get back on the results.

Any thoughts on how well DIRAC could be able to do this job?
I do like the idea that it takes multiple measurements in to account.

You can try to EQ each subwoofer separately for more precise control.
Start by measuring the response of your entire audio system after applying initial EQ settings.
Next, EQ each subwoofer individually, generate unique filters for each, and apply them separately.
Afterward, remeasure the complete system and compare the results to the initial measurement.
Additionally, trust your ears—listen carefully to the differences to determine which configuration sounds best to you.

However, in my opinion this additional control may overcomplicate things and make the result worse. For example, I tried to use MSO (Multiple Sub Optimizer) to calculate delays and different EQs and gains, separate for each channel, and the result was much worse than what I got with REW, applying the same PEQ only to the subs.

In my opinion, Dirac feels like a project a computer science student might complete over a summer, whereas REW (Room EQ Wizard) is far more advanced. Dirac can be effective, but its features are limited, offering roughly like 5% of REW’s capabilities.

 

[WeekendWarrior]

I used the right calibration file (90 degrees) and pointed the microphone upwards

I thought the correct cal file was 0 degrees with the mic pointed towards the speakers. The 90 deg cal file with mic pointed upwards is for surround sound?

Assuming you have a UMIK mic of course.

EDIT: Or perhaps if you have subs distributed around the room rather than next to your main speakers it is correct to measure it more like a surround sound set-up?

 

[Geertidow]

On the left side, both your speaker and subwoofer exhibit a null at 69 Hz. However, this shouldn't be a major concern, as the right side will partially compensate for it.

Here’s my suggested approach. It’s not necessarily the definitive solution, nor does it guarantee you’ll prefer it, but it’s what I would try first.

I wouldn’t apply EQ to the R3s (except setting crossover frequency on them), as their frequency response is already quite good. Instead, I’d focus on EQing the subwoofers only.

You can keep the crossover frequency at 66 Hz for both the subwoofers and speakers. Personally, I prefer a slightly lower crossover frequency, around 58–60 Hz, since my speakers handle frequencies above 30 Hz well. The rationale is that many subwoofer enthusiasts, myself included, often boost the bass. If the crossover frequency is set at 70 Hz or higher (or even slightly below), this boost can affect the 70–80 Hz range, which often sounds "boomy." A lower crossover helps avoid this issue.

Additionally, I recommend using a steep filter curve, such as a Butterworth 48 dB/octave. This ensures that the subwoofers and speakers only reproduce their intended frequency ranges. A shallow filter curve allows some unwanted frequencies to bleed through. For example, setting the crossover to 58–60 Hz with a 48 dB/octave curve effectively suppresses frequencies around 70 Hz and above.

Here’s how your crossover settings should look:

View attachment 478217
But again, personally, I would set the crossover frequency lower, to 58–60 Hz.

Next, I opened your "Total system, no Dirac" measurement in REW and accessed the EQ window. I set the target level to 90 dB, which is significantly higher than the estimated or average level. I assumed you might want to boost the bass, as many enthusiasts, including myself, enjoy this enhancement.

I then generated the filters using the "Match response to target" option and saved them to a file ("Save filter coefficients to a file", selecting miniDSP and 48 kHz in the popup window).

Here’s the generated filter file: http://ug.link/pashka/filemgr/share-download/?id=ce3d17c2130a4fcebc586d1a37fb2162

In miniDSP, select "PEQ" for your first subwoofer and lick the "Menu" button, then link your first sub to the second sub. Then, choose "Load Biquads File," and upload the generated filter file linked above.
You should see the EQ graph change according to the filters.
View attachment 478218
You might be satisfied with the results as they are. If you want to increase the bass further, simply adjust the gain on your subwoofers, like I often do:
View attachment 478219

Let me know if you like the results.

So after some listening using a crossover with a steep slope really helped (I think). Also DIRAC handles it much, much better when a steep crossover is applied.
I've included some measurements with and without DIRAC live (left channel, right channel and full range).
As a comparison, I've attached manual PEQ using REW (only full range).
Each channel consists of KEF R3 meta + SVS pro 2000.

Indeed a steep crossover seems to make the MiniDSP handel the DIRAC well.
Link to mdat: wetransfer link (valid 3 days)

 

[Geertidow]

I thought the correct cal file was 0 degrees with the mic pointed towards the speakers. The 90 deg cal file with mic pointed upwards is for surround sound?

Assuming you have a UMIK mic of course.

O wow that would be something, I always thought 90 degrees was upward, no idea why.
It does not seem to matter all that much, but I'll redo DIRAC if that's really the case.

 

[ViperDom]

@Geertidow, recently I asked someone to share their opinion on KEF R3s with subwoofers. Maybe it was you, but if not, can you please share how well they work together? Because I already have subwoofers and thinking of upgrading some of my speakers to KEF R3s.

IMO, with a multi-sub setup you have the option to save the R3$ and go with the LS50 Meta's. I run these with Dual OutlawAudioX12's processed with minidsp&dirac. Very much enjoy the precise imaging and optimized frequency response. I just set the DIRAC target curve with a typical Harman slope and then adjust bass boost via minidsp shelf filter on-the-fly to taste.

I've experimented with ported vs plugged (measured & listened) and didnt hear any difference once you slap that 48db X-over on em. I'm not sure which way(port/plug) is technically 'better' but decided to forgetabout it and just enjoy the music.
Regarding the OP, perhaps he plugged the ports in attempt to 'match' the slope his sealed subwoofers, which is a 'thing' in the absence of miniDSP's super custom crossover options.

 

[WeekendWarrior]

O wow that would be something, I always thought 90 degrees was upward, no idea why.
It does not seem to matter all that much, but I'll redo DIRAC if that's really the case.

Well, that's what I thought. Happy to be corrected. And perhaps the edit to my last post is relevant too?

 

[Geertidow]

You can try to EQ each subwoofer separately for more precise control.
Start by measuring the response of your entire audio system after applying initial EQ settings.
Next, EQ each subwoofer individually, generate unique filters for each, and apply them separately.
Afterward, remeasure the complete system and compare the results to the initial measurement.
Additionally, trust your ears—listen carefully to the differences to determine which configuration sounds best to you.

However, in my opinion this additional control may overcomplicate things and make the result worse. For example, I tried to use MSO (Multiple Sub Optimizer) to calculate delays and different EQs and gains, separate for each channel, and the result was much worse than what I got with REW, applying the same PEQ only to the subs.

In my opinion, Dirac feels like a project a computer science student might complete over a summer, whereas REW (Room EQ Wizard) is far more advanced. Dirac can be effective, but its features are limited, offering roughly like 5% of REW’s capabilities.

I did this, I tinkered a bit and corrected up to 300 hz.
Currently comparing that (I took care of level matching after correction) and included the results in the mdat data in the abovementioned post (link mdat, it's only 3 days valid).
Both routes work well, but now I'm listening and switching.

Both sound really nice so far, and the low bass is still there
But, @OCA maybe supplying some really great other crossovers and filters, so who knows.
The crossover at 48 db/oct really helped for the DIRAC implementation though!

 

[Geertidow]

IMO, with a multi-sub setup you have the option to save the R3$ and go with the LS50 Meta's. I run these with Dual OutlawAudioX12's processed with minidsp&dirac. Very much enjoy the precise imaging and optimized frequency response. I just set the DIRAC target curve with a typical Harman slope and then adjust bass boost via minidsp shelf filter on-the-fly to taste.

I've experimented with ported vs plugged (measured & listened) and didnt hear any difference once you slap that 48db X-over on em. I'm not sure which way(port/plug) is technically 'better' but decided to forgetabout it and just enjoy the music.
Regarding the OP, perhaps he plugged the ports in attempt to 'match' the slope his sealed subwoofers, which is a 'thing' in the absence of miniDSP's super custom crossover options.

I plugged the ports of the KEF R3 (the front of the speakers are 50 cm from the wall), I think if I look at the measurements it does a fine job.
I thought of the LS50 meta's but I like the flexibility of the KEF R3M, think they are gorgeous and enjoy lower crossovers (if possible).
But of course would've been happy with the LS50 meta.

 

[Geertidow]

View attachment 478346

Well, that's what I thought. Happy to be corrected. And perhaps the edit to my last post is relevant too?

I point upward because once someone said to just always do that.
There is no more logic behind it than that.

 

[PashkaTLT]

I thought the correct cal file was 0 degrees with the mic pointed towards the speakers. The 90 deg cal file with mic pointed upwards is for surround sound?

Assuming you have a UMIK mic of course.

EDIT: Or perhaps if you have subs distributed around the room rather than next to your main speakers it is correct to measure it more like a surround sound set-up?

Here's what miniDSP say on their website:

• For mono or stereo systems, use the 0-degree file and point the UMIK-1 directly at the speakers.
• For multichannel systems (e.g., 5.1 or 7.1) or surround setups with speakers spread around the room, use the 90-degree file and point the UMIK-1 at the ceiling.

 

[PashkaTLT]

So after some listening using a crossover with a steep slope really helped (I think). Also DIRAC handles it much, much better when a steep crossover is applied.
I've included some measurements with and without DIRAC live (left channel, right channel and full range).
As a comparison, I've attached manual PEQ using REW (only full range).
Each channel consists of KEF R3 meta + SVS pro 2000.

Indeed a steep crossover seems to make the MiniDSP handel the DIRAC well.
Link to mdat: wetransfer link (valid 3 days)

Glad you're getting improvement!
Do you think it's because the steep crossover removes those boomy frequencies more effectively?

 

[PashkaTLT]

IMO, with a multi-sub setup you have the option to save the R3$ and go with the LS50 Meta's. I run these with Dual OutlawAudioX12's processed with minidsp&dirac. Very much enjoy the precise imaging and optimized frequency response. I just set the DIRAC target curve with a typical Harman slope and then adjust bass boost via minidsp shelf filter on-the-fly to taste.

I've experimented with ported vs plugged (measured & listened) and didnt hear any difference once you slap that 48db X-over on em. I'm not sure which way(port/plug) is technically 'better' but decided to forgetabout it and just enjoy the music.
Regarding the OP, perhaps he plugged the ports in attempt to 'match' the slope his sealed subwoofers, which is a 'thing' in the absence of miniDSP's super custom crossover options.

I thought so too, @ViperDom, I have 5" speakers in my HT with 12" sub (2nd 12" sub in the process of DIYing ) and I enjoy the sound (as much as Klipsch can be enjoyed, because of course their budget and mid-range speakers and subs can't compete with serious expensive options from other manufacturers).

But I created this thread on this very subject here:

'Do I need 8" woofers in my speakers if I'm going to have 2 subwoofers?'

Sep 16, 2025

Hello,

I'm choosing future speakers, perhaps for my home theater or perhaps for my music system (if new speakers beat my current KRK Rokit 10-3).
My HT will have 2 x 12" subwoofers and my music system has 2 x 18" subwoofers.

I will be getting some really good 3-way speakers (looking at KEF R3 currently) and my question is: do I need to get speakers with large (~8") woofers or having 5-6.5" woofers will be absolutely enough provided I will have subwoofers?

• PashkaTLT
• Replies: 27
• Forum: Speaker Reviews, Measurements and Discussion

And the general consensus is that even with subs, the main speakers should have at least 6.5" woofers. The reasons are, iirc:
- Speakers have more distortions closer to their range limits, especially on higher volume. So the speaker can play 60 Hz, for example, but the distortions at that frequency can be higher than at 200 Hz, for example, at least on higher volumes.
- Speakers with bigger woofers are designed to handle higher SPL better

 

[PeterNL]

So after some listening using a crossover with a steep slope really helped (I think). Also DIRAC handles it much, much better when a steep crossover is applied.
I've included some measurements with and without DIRAC live (left channel, right channel and full range).
As a comparison, I've attached manual PEQ using REW (only full range).
Each channel consists of KEF R3 meta + SVS pro 2000.

Indeed a steep crossover seems to make the MiniDSP handel the DIRAC well.
Link to mdat: wetransfer link (valid 3 days)

Looks better, much lower distortions (previously were sky high), still...your room is very reverberant. How does it sound ??

 

[Geertidow]

Glad you're getting improvement!
Do you think it's because the steep crossover removes those boomy frequencies more effectively?

It is also something with the DIRAC itself, but dont know what. Or my preference is just weird. I added 2 dB's for the subs on the DIRAC setting.

 

[PashkaTLT]

View attachment 478346

Well, that's what I thought. Happy to be corrected. And perhaps the edit to my last post is relevant too?

Glad to see a fellow AI enjoyer.

If anyone is interested in AI,
I do AI battles as a hobby and in my last AI tournament Grok 3 was on the 1st place, with ChatGpt-4o (through API or MS Bing Copilot) and Gemini on the 2nd place. I haven't tested Grok 4 & Chatgpt-5 yet, though.
Also recently I installed the new Comet AI browser from the Perplexity team and it can do some really advanced stuff and help save much time.