Dirac Live Bass Control Multi SW - seeking info

[Mike-48]

I am trying to understand just what this does and also what hardware it supports.

From the very little I found on the Dirac Web search, I've inferred that perhaps it uses a random or guided random search to find level, phase, polarity settings (as well as EQ settings) that optimize smoothness of multiple subwoofers. Is that right? For example, if SW1 has a null at 40 Hz, might it increase output of the other SWs there to compensate? Would it search for phase settings on multiple SWs to optimize combined output at a listening position or positions?

Would it support, e.g., the NAD M66 with its four SW outputs?

 

[JStewart]

The goal of DLBC is to minimize naturally occurring bass response variation (in listening rooms)for multiple listening positions and to integrate the subs and speaker groups response together, i.e. subs and mains, subs and surrounds, and so on.

Dirac is a “black box” solution so no one can precisely say what it’s doing. Safe to assume your inferences are on the right track though. Spending some time with MSO, and the recommended reading list found there, can be a very educational experience as to more precisely how such things work.

Multi-Sub Optimizer Freeware for Multiple Subwoofers

Optimize bass performance of multiple-subwoofer audio systems. Reduce seat-to-seat frequency response variations.

www.andyc.diy-audio-engineering.org

As for DLBC support, look for specific information on any separate piece of equipment for support confirmation. On NAD’s website for the M66, support for DLBC is stated.

I started with MSO before DLBC was available. I use DLBC now because it greatly simplifies things when something in the room changes and recalibration needs done. Also one less piece of external hardware is needed.

If you’re in the market for a piece of gear that can run DLBC and can wait, perhaps a year or two, Dirac Active Room Treatment (ART) will likely become available for new high end two channel processors, like the M66.
At the moment, and through the 3rd qtr of this year, the only pieces of gear running ART are from Storm Audio.

Thread 'Game Changer! Dirac Officially Unveils Active Room Treatment (ART) Technology'
https://www.avnirvana.com/threads/g...s-active-room-treatment-art-technology.11545/

StormAudio

Immersive audio technologies to High End Home Theaters

www.stormaudio.com

 

[Mike-48]

Thanks, @JStewart. I wish Dirac were a little more clear about exactly what is being optimized.

ART looks interesting, too. My impression of that one is that it uses the speakers for active cancellation of excess reverberation.

It does seem that conventional approaches today -- by which I mean algorithms that optimize each speaker individually and stop there or possibly add optimization of phase of mains and corresponding subs -- are just scratching the surface of what's imaginable in bass optimization.

 

[JStewart]

To be fair, I think Dirac’s intent is an automated system that works without the need for any understanding, which will be a much larger chunk of this very small market. It works extremely well in that regard. The problem Dirac is running into a.t.m. is buggy implemented by hardware manufacturers.

ART, as I understand it, does use the speakers in a system to support each other speaker and to actively cancel reflections that cause nulls or peaks in SPL. While automated, there are presently use cases where user intervention is required and the user needs a good understanding of what they are doing. As long as this remains the case I suspect ART will remain in the realms of the custom installers and enthusiasts.

I failed to mention that DLBC also handles the crossovers, which are user adjustable, and ART doesn’t use crossovers in the traditional sense, but does handle the transition from subwoofers to the other speakers.

 

[ddude003]

The last paragraph at https://www.dirac.com/live/bass-control/ sums it up pretty well...

"Dirac Live Bass Control leverages all-pass filters, plus machine learning and artificial intelligence, to calculate gains, delays, and all-pass filters for each subwoofer. In this way, the Dirac solution ensures that low frequencies add up so that not only the average is controlled, but most importantly, seat-to-seat variation is minimized as we can now achieve a level of control in multiple locations at once."

I don't think the folks at Dirac are going to give away much more of their secret sauce than this... Lets hope the AI is smart enough to not boost nulls...

 

[mitchjrj]

I failed to mention that DLBC also handles the crossovers, which are user adjustable...

Hey J

Failing to find an answer perhaps you can offer some insight. Just loaded DLBC for my Marantz Cinema 50 today and going through the additional features in the software. I'm confused about crossovers; on the one hand the DLBC documentation says it "guides" the user in their crossover selection, rather than outright saying they identify the optimal XO point. Not very helpful considering any XO adjustments do not render in real time and require re-calculation each iteration.

The initial calculation suggests 70Hz for my mains/center, 72Hz for side surrounds, and 75Hz for rear surrounds.

Here are my questions...

1) How do DLBC's crossover recommendations translate to an AVR with the typical steps of 40/60/80/90/100Hz (etc)? Do I select the next highest option (ie. 80Hz)? Alternative?
2) Were DLBC's crossover recommendations actually calculated and optimized for my room and equipment? Or are they generic starting points? I saw a DLBC video where the author indicated 70Hz was his default.

I haven't loaded the correction to my AVR yet, don't know if DLBC has the capability to override the AVR's crossover points and enter whatever you set. Any clarity you can provide here would be greatly appreciated.

 

[mitchjrj]

How do DLBC's crossover recommendations translate to an AVR with the typical steps of 40/60/80/90/100Hz (etc)?

Loaded the new settings, answered part of my question—indeed DLBC replaces the AVR’s crossover points with their own (or whatever you selected).

Now I need to understand what informed DLBC’s “recommended” crossovers, and would inspire me to set them otherwise.

 

[Sonnie]

ART, as I understand it, does use the speakers in a system to support each other speaker and to actively cancel reflections that cause nulls or peaks in SPL. While automated, there are presently use cases where user intervention is required and the user needs a good understanding of what they are doing. As long as this remains the case I suspect ART will remain in the realms of the custom installers and enthusiasts.

Adding to this, I'm in the middle of reviewing the StormAudio MK3 with DLART as well as writing somewhat of a setup guide within the review. In addition to helping with reflections, the user must tell EACH set of speakers how it will help support other speakers in the room. You could have every speaker capable of playing from 20Hz to 150Hz to help any speaker not capable of playing any part of that frequency range. There are no set crossovers to speak of, but you may tell your rear speakers to help the side surrounds from 50-150Hz, but only help the front speakers from 50-100Hz, then tell the rear subs to help any speaker needing help from 20-80Hz.. maybe tell your front subs to help the front speakers from 20-150Hz, but only help the rear speakers from 20-80Hz. Maybe some speakers don't help any of the others. Every speaker can help support every other speaker in the frequency range the speaker helping is capable of supporting and at a level from -1dB to -24dB (higher is more help). It's a lot of settings to work on, so it's a highly time-consuming process. I can play a strictly stereo song and turn off the two main front speakers and hear what all the support speakers are playing; it's quite an odd sound, and I can view the speaker output levels of those that are helping from the software as well. Very sophisticated stuff going on with DLART... and you are right... the implementation is very important, which is probably why it won't be an overnight inclusion of DLART in other processors... it will take time for sure.

 

[JStewart]

1) How do DLBC's crossover recommendations translate to an AVR with the typical steps of 40/60/80/90/100Hz (etc)? Do I select the next highest option (ie. 80Hz)? Alternative?
2) Were DLBC's crossover recommendations actually calculated and optimized for my room and equipment? Or are they generic starting points? I saw a DLBC video where the author indicated 70Hz was his default.

1. As you already figured out DLBC crossover filters are used, not the AVR’s.

2. DLBC doesn’t select nor claim to select the best crossover point. It will pick the lowest point it thinks the speaker group is capable of if above 70Hz and 70Hz otherwise.
When filters are calculated DLBC will show the response at the XO. If it’s not satisfactory, i.e. following the chosen target curve, you can recalculate as the result may be different, or choose a different XO and recalculate. Check all speaker groups.

Use this guide to select a starting XO value:

Setting crossovers in an AVR/pre-pro with Dirac Live 2 room correction

The XO guide is from a 3rd party and recommended by Dirac themselves in their FAQ found on this page.

Helpdesk - Dirac

Dirac Live Support Knowledge base & Support Dirac Live official calibration guide. Dirac optimizes digital audio, perfecting sound for better listening in any environment. See...

www.dirac.com

Lots of other good info/resources can be found there other than the bad YouTube calibration video which doesn‘t show correct mic handling and measurement technique. It also contradicts accurate information Dirac provides elsewhere.

 

[mitchjrj]

Use this guide to select a starting XO value:

Setting crossovers in an AVR/pre-pro with Dirac Live 2 room correction

The XO guide is from a 3rd party and recommended by Dirac themselves in their FAQ found on this page.

Helpdesk - Dirac

Dirac Live Support Knowledge base & Support Dirac Live official calibration guide. Dirac optimizes digital audio, perfecting sound for better listening in any environment. See...

www.dirac.com

That is a good guide. Using that approach—assessing falloff points—I ended up with some wildly different values than the defaults. Haven’t had time to test them, or use REW to document actual response.

I still wonder if it’s worth the effort to chose an escalating series of XO’s in DLBC (70, 75, 80 etc), recalculate, upload, measure each. Then compare the resulting REW frequency response curves to eyeball the smoothest response. Fine tune from there.

 

[JStewart]

Using that approach—assessing falloff points—I ended up with some wildly different values than the defaults.

Yup, ver few of us have the very same speaker in all positions.

I still wonder if it’s worth the effort to chose an escalating series of XO’s in DLBC (70, 75, 80 etc), recalculate, upload, measure each. Then compare the resulting REW frequency response curves to eyeball the smoothest response. Fine tune from there.

The Dirac calculation is accurate, however it’s accurate for the average of the measurements and not the single measurement for the MLP that you’ll get with REW. If you prefer to optimize for the MLP use the 9 point measurement pattern as a start. If you want to test if Dirac works just take a single measurement with Dirac and upload a filter from it alone. Then without having moved the mic take a REW measurement.
By the way, if you do upload a filter from a single measurement you can check the R and L delays to be sure the mic is exactly centered if that applies to your setup. If you want to center the mic a little more quickly then turn off all speakers other than R and L. The measurement and upload will be much faster.
If you want to experiment with different mic patterns then after you take the first centered measurement save the project with just that measurement before and without proceeding to the filter creation screen. You can then load this project as your starting point for future projects with different patterns.

 

[mitchjrj]

If you want to test if Dirac works just take a single measurement with Dirac and upload a filter from it alone. Then without having moved the mic take a REW measurement.

If you want to experiment with different mic patterns then after you take the first centered measurement save the project with just that measurement before and without proceeding to the filter creation screen. You can then load this project as your starting point for future projects with different patterns.

Wild info, thanks!

 

[Mike-48]

The question I still have about DLBC is this: If there are four subs, does it do its best with each of them independently and that's it, or is there some attempt to adjust the sound from all four playing at once?

It could be that sub A has a dip at (say) 40 Hz, and sub B has plenty of output. Would the surplus in B be used to make up for the dip in A? That won't happen if individual adjustment is all it does.

Will it adjust the phase of the subs to get the smoothest overall response?

It's simple math to get one sub equalized to sound right, but it's a lot more complicated for four, given all the phase and amplitude interactions. Just what is this software promising?

 

[Sonnie]

The question I still have about DLBC is this: If there are four subs, does it do its best with each of them independently and that's it, or is there some attempt to adjust the sound from all four playing at once?

It could be that sub A has a dip at (say) 40 Hz, and sub B has plenty of output. Would the surplus in B be used to make up for the dip in A? That won't happen if individual adjustment is all it does.

Will it adjust the phase of the subs to get the smoothest overall response?

It's simple math to get one sub equalized to sound right, but it's a lot more complicated for four, given all the phase and amplitude interactions. Just what is this software promising?

It will set phase and gain, and your response of all subs combined should be very similar, if not an almost exact match, to your target curve.

As previously mentioned... test it... take one or multiple measurements at the MLP without moving the mic, then measure it with REW and see your response. When you measure, the sweep will play thru all subs combined.

The only way it can be acceptable to anyone will be to have all subs tuned to play at the same time, as that's how they play.

 

[Mike-48]

The only way it can be acceptable to anyone will be to have all subs tuned to play at the same time, as that's how they play.

Yup. The promotional literature is quite unclear that it does it that way, so I'm glad for clarification. For example, the promo for the new NAD M66 says

In a first for a stereo component, the M66 features four balanced and four single-ended subwoofer outputs and integrates Dirac Live Bass Control, which allows independent calibration of multiple subwoofers.

The word "independent" can be taken several ways, and it me it wasn't obvious that it can calibrate the four together, not just one at a time.

 

[Sonnie]

Remarkably, it can take each sub's measurements independently and combine them for a smooth combined response without measuring them combined.

Back when we were dialing in subs with the BFD processor, we had to measure the subs combined and add PEQ to all subs simultaneously. When we tried to PEQ each sub and then combine them... the response was all over the place. Technology has come a long way since those days.

 

[Mike-48]

Yes, combining subs is hardly trivial!