New Dirac MIMO active room correction

[Sancus]

The Storm Audio announcement includes the word "exclusive" so it may be locked out of any other platform for a time. I hope that's not true, though.

I also would like to see a comment from @Flak

 

[stevenswall]

The Storm processors, currently the only ones capable of the new functions, have more processors than most mainstream AVR's - 4 vs 2.

Given what Dirac are trying to do with this new capability, I would expect it to be processor intensive, and I would not find it surprising if current/past processors/AVR's weren't up to it. Sure marketing will come into it, but this is an order of magnitude shift in the amount of processing needed, even if you allow an increase in lag... at some point, you never catch up, you just fall further and further behind (unless you drop frames/data to catch up).

Most of the AVR's are based on a single SOC chipset from analog devices SHARC family - these max out at dual processor - and are the norm - with devices from Onkyo, Denon and others supporting Dirac Live, Audyssey, and various other functions without any trouble.

The StormAudio ISP runs quad SHARC processors - so at least two dual CPU analog devices SOC's - they also use ARM processors for the control systems (Analog Devices also have combo chipsets with embedded ARM processor in addition to 2xSHARC processors - https://www.analog.com/en/products/adsp-sc592.html#product-overview - although last I heard these were in pre-release... hard to tell precisely what chipsets StormAudio are using)

At an indicative minimum - these StormAudio processors have double the processing power of most current Dirac capable devices - perhaps there is a good reason for this, and we are seeing it now.

Do you know how fast these processors are in laymans terms?

Seems like if it's processing stuff in real time, they could afford to put a high end CPU in rather than an embedded devices one, or if it's just compute intensive for the initial calculation, they could do what Genelec does and offload it to the cloud.

If it's the latter, there's no excuse.

 

[abdo123]

Wont we need fullrange surrounds to make this work?

Yes you do. Or at least something like the Arendal 1961 that can pull their weight because of their sealed configuration.

 

[tifune]

Yes you do. Or at least something like the Arendal 1961 that can pull their weight because of their sealed configuration.

Can you elaborate a bit? I'm mostly wondering what criteria need to be met for this solution to work as advertised

 

[abdo123]

Can you elaborate a bit? I'm mostly wondering what criteria need to be met for this solution to work as advertised

Basically you need as many speakers that can play low frequencies as you can to provide the algorithm with the most possible amount of options with regards to which speaker to utilise to cancel out which peak or dip.

As a result you’re also SPL limited by the weakest speaker in your System.

Increasing number of speakers also allows the algorithm to reduce ringing higher up in frequency.

Audioholics made an extensive video about it.

 

[Axo1989]

Haha, Gene calls them DEE-rack. I bet Erin does the same. Definitely watching the rest of that when I have some time, looks interesting.

 

[Keened]

Basically you need as many speakers that can play low frequencies as you can to provide the algorithm with the most possible amount of options with regards to which speaker to utilise to cancel out which peak or dip.

I think the important caveat to mention here is that one can get extremely good results with 2-3 spatially separated speakers for the 10-200hz range already, so two sufficiently large mains and a single subwoofer should be enough (or 2-3 subwoofers and two mains). The higher frequencies take less energy so it's not like the satellites have to be massive if they're asked to only correct/support for anything higher than that..

If you're going to try to have a ruler flat result over several rows, well you should already be using 4+ subwoofers anyways.

-----------------------------

Watching the video, nothing I've seen so far seems to contradict what I've heard. Not going to help much for people who only have front speakers with listening positions up against a wall though

Edit: unless someone introduces a series of super thin 12" speakers exclusively used to create counter-pressure for the area immediately in front of them...

 

[Music707]

The Storm Audio announcement includes the word "exclusive" so it may be locked out of any other platform for a time. I hope that's not true, though.

I also would like to see a comment from @Flak

Well, the text says "exclusive partner in the introduction of Dirac Live Active Room Treatment" which literally is just right now. Other manufacturers may follow sooner or later.. Given that Storn Audio sells its processors at a high price point sales numbers may be relatively small and the Dirac people might even count on additional partners in order to justify the R&D cost and effort for Dirac ART.

 

[kemmler3D]

Given what Dirac are trying to do with this new capability, I would expect it to be processor intensive,

Probably, but not not severely so once it's set up. I mean, hell, a pair of headphones with a $8 chip can do "advanced" ANC and this a similar principle. Especially given that it's meant for low frequencies only, I don't see why the CPU load should be super high.

I think what would be happening is the processor will be running N^2 FIRs where N is the number of speakers. Each speaker will be emitting a filtered version of each other speaker's signal (in addition to the original signal) in order to produce the right cancellations and response at the MLP.

Now, it could be more complex or completely different than that, that's just my vague guess at how this works.

exclusive partner in the introduction of

As a marketing professional, I would say the phrase "in the introduction of" means they have a short agreement to exclusivity, if any. It is a phrase I would use to pump up the fact that nobody else happens to be launching products with the feature at the moment, so it's technically "exclusive".

More likely, they mutually agreed to work on the initial implementation because it's a fairly innovative / new product and they would need to work closely to iron out kinks and run support together. However, this doesn't mean Dirac agreed to withhold the tech from other partners after the initial launch is done with. As such I'd GUESS we'll see other implementations as soon as 5-6 months after launch if things go very smoothly, probably more like 12-18.

Audioholics made an extensive video about it.

Is my vague guessing about how the system works in any way accurate?

Anyway, I am pretty interested in this tech, it finally promises something you can't really do with EQAPO and sheer determination. I imagine that if you have a few subs, it would probably benefit the sound more than an additional sub. In which case you'd have to say it's worth an extra $500-1500 or so?

 

[dlaloum]

Do you know how fast these processors are in laymans terms?

Seems like if it's processing stuff in real time, they could afford to put a high end CPU in rather than an embedded devices one, or if it's just compute intensive for the initial calculation, they could do what Genelec does and offload it to the cloud.

If it's the latter, there's no excuse.

The analog devices specs say 1ghz or 800mhz depending on model.

 

[poxymoron]

This looks very interesting. Hope it comes to the miniDSP Flex eventually.

 

[Ata]

Yes you do. Or at least something like the Arendal 1961 that can pull their weight because of their sealed configuration.

Since most speakers, even smaller bookshelves, can do 80Hz at reasonable SPL, and the room transition frequency is often at a frequency much higher than that, especially in smaller rooms, I expect that full-range speakers all around should be nice to have, but not must have for the operation of DIRAC MIMO. Small speakers can fix issues between their -3db freq and room frequency, up to a certain SPL.

 

[abdo123]

Since most speakers, even smaller bookshelves, can do 80Hz at reasonable SPL, and the room transition frequency is often at a frequency much higher than that, especially in smaller rooms, I expect that full-range speakers all around should be nice to have, but not must have for the operation of DIRAC MIMO. Small speakers can fix issues between their -3db freq and room frequency, up to a certain SPL.

In my experience it’s absolutely horrendous to be sitting in the middle of a movie fully immersed and then at the peak of the movie you start hearing some of your speakers farting.

I expect this to be even worse because the fundemental would not be audible but the distortion/farting will be.

So yeah having more sources is definitely better than not having any sources at all. But at least pick sealed speakers meant for home theatre applications instead of super tiny ported ones that will inevitably start farting at some point.

 

[dlaloum]

There has been an interview with Dirac by AVNirvana:

Game Changer! Dirac Officially Unveils Active Room Treatment (ART) Technology

(January 5, 2023). Just a few short days into the new year and Dirac, makers of the popular Dirac Live Room Correction suite, has launched a game-changing home audio technology. Its new “Active Room Treatment” or ART delivers a layer of room correction designed to hone and control frequencies up...

www.avnirvana.com

Items of interest:

1. the new Dirac Live extension is being called ART (Active Room Treatment) - so now we have Dirac Live, DLBC, DLART
2. Although in a controlled environment (like a car) ART is capable of being used up to 4kHz, initial released version will be limited to 150Hz, although they think that they will fairly quickly extend this to 300Hz. - This is particularly sad - I was hoping to see something up to 1kHz... but apparently not this year!
3. Initial release will be on StormAudio processors in Spring (Q2) - with others likely Q4 (?)
4. Although there is a performance impost due to the additional functionality, it is not massive (as I previously believed it would be) - and it was specifically stated that devices capable of DLBC would be able to run ART. There was some talk that even base Dirac Live devices might be able to run ART... this seemed somewhat less firm, in the way it was said. (would be nice to see a DL-ART upgrade for Onkyo RZ50, or my Integra DRX3.4!)
5. Reality is that we are unlikely to see this in mainstream processors till at least late this year - it will initially be reserved for StormAudio and the stratospheric $$$ end of the market.
6. StormAudio CEO stated that the upgrade for existing Dirac models to Dirac-ART will be $299 - that is an upgrade from full Dirac with multisub DLBC - that provides an indication of likely upgrade prices...
7. It is early days, and Dirac CEO clearly states that decisions with regards to marketing tiers and pricing has not been made - so there may be a future in which one could purchase more limited frequency DL-ART at a lower price, and more extended frequency DL-ART for more $$$.

P.S. Gene at Audioholics is apparently reviewing a StormAudio processor with DL-ART.... pending, I presume, NDA's...

 

[Ata]

In my experience it’s absolutely horrendous to be sitting in the middle of a movie fully immersed and then at the peak of the movie you start hearing some of your speakers farting.

I expect this to be even worse because the fundemental would not be audible but the distortion/farting will be.

So yeah having more sources is definitely better than not having any sources at all. But at least pick sealed speakers meant for home theatre applications instead of super tiny ported ones that will inevitably start farting at some point.

"Farting" is a problem only if bass management is not done well. My assumption is that the DIRAC MIMO aka ART is smart enough to bass manage the small(er) speakers so as to not physically overwhelm them up to certain (reference?) SPL (and maybe above that SPL, if the bass management is done in a volume aware fashion, e.g. to limit the bass as some active speakers do).

In other words, given high enough SPL at low enough frequency, any non bass managed speaker, including the very expensive full range ones, will go into non-linear behaviour. The only thing that changes from speaker to speaker is what SPL and what frequency. That was my point.

 

[Ata]

The Storm processors, currently the only ones capable of the new functions, have more processors than most mainstream AVR's - 4 vs 2.

Given what Dirac are trying to do with this new capability, I would expect it to be processor intensive, and I would not find it surprising if current/past processors/AVR's weren't up to it. Sure marketing will come into it, but this is an order of magnitude shift in the amount of processing needed, even if you allow an increase in lag... at some point, you never catch up, you just fall further and further behind (unless you drop frames/data to catch up).

Most of the AVR's are based on a single SOC chipset from analog devices SHARC family - these max out at dual processor - and are the norm - with devices from Onkyo, Denon and others supporting Dirac Live, Audyssey, and various other functions without any trouble.

The StormAudio ISP runs quad SHARC processors - so at least two dual CPU analog devices SOC's - they also use ARM processors for the control systems (Analog Devices also have combo chipsets with embedded ARM processor in addition to 2xSHARC processors - https://www.analog.com/en/products/adsp-sc592.html#product-overview - although last I heard these were in pre-release... hard to tell precisely what chipsets StormAudio are using)

At an indicative minimum - these StormAudio processors have double the processing power of most current Dirac capable devices - perhaps there is a good reason for this, and we are seeing it now.

My assumption, possibly wrong one, is that, ultimately, any multi-channel DRC should come down to some per channel FIR convolution, which most DIRAC and MultEQ XT32 receivers can already do at some OK number of taps.

The front-end part, where these FIR filters need to be computed and optimised, can be infinitely compute expensive, and that is OK, because it is all done offline, upfront, and in the cloud where compute power is not a problem.

 

[abdo123]

"Farting" is a problem only if bass management is not done well. My assumption is that the DIRAC MIMO aka ART is smart enough to bass manage the small(er) speakers so as to not physically overwhelm them up to certain (reference?) SPL (and maybe above that SPL, if the bass management is done in a volume aware fashion, e.g. to limit the bass as some active speakers do).

In other words, given high enough SPL at low enough frequency, any non bass managed speaker, including the very expensive full range ones, will go into non-linear behaviour. The only thing that changes from speaker to speaker is what SPL and what frequency. That was my point.

That’s actually a very important point, finding the sweetspot which in this case is the crossover frequency seems impossible without some sort of human intervention. (Knowing which speaker’s port is tuned to which frequency .etc)

From my experience automatic bass management overestimate a speaker’s low frequency capability because room gain and room modes messes the -3dB and -10 dB point of the speaker.

 

[Ata]

That’s actually a very important point, finding the sweetspot which in this case is the crossover frequency seems impossible without some sort of human intervention. (Knowing which speaker’s port is tuned to which frequency .etc)

From my experience automatic bass management overestimate a speaker’s low frequency capability because room gain and room modes messes the -3dB and -10 dB point of the speaker.

Yes, it can be tricky for fully automated systems, but if one goes with the typical for AVR Fxo of 80 or 100 Hz for small speakers, port and bass THD effects should be largely eliminated. Of course, that will mean limited MIMO effect under Fxo, but full MIMO above Fxo to at least room transition frequency and hopefully another octave above that.

 

[Dj7675]

It is interesting how long this took for Dirac to get this out. It seems to have had many names over the years. I have a Stormaudio processor with 7 JBL 6332 speakers. It will be interesting to play with it once it is out. It is also interesting that they are starting with the limit of 150hz. I wonder if there will be a curtain like with Dirac at some point where you set how high up in frequency you want to use ART.

 

[Everett T]

The Storm processors, currently the only ones capable of the new functions, have more processors than most mainstream AVR's - 4 vs 2.

Given what Dirac are trying to do with this new capability, I would expect it to be processor intensive, and I would not find it surprising if current/past processors/AVR's weren't up to it. Sure marketing will come into it, but this is an order of magnitude shift in the amount of processing needed, even if you allow an increase in lag... at some point, you never catch up, you just fall further and further behind (unless you drop frames/data to catch up).

Most of the AVR's are based on a single SOC chipset from analog devices SHARC family - these max out at dual processor - and are the norm - with devices from Onkyo, Denon and others supporting Dirac Live, Audyssey, and various other functions without any trouble.

The StormAudio ISP runs quad SHARC processors - so at least two dual CPU analog devices SOC's - they also use ARM processors for the control systems (Analog Devices also have combo chipsets with embedded ARM processor in addition to 2xSHARC processors - https://www.analog.com/en/products/adsp-sc592.html#product-overview - although last I heard these were in pre-release... hard to tell precisely what chipsets StormAudio are using)

At an indicative minimum - these StormAudio processors have double the processing power of most current Dirac capable devices - perhaps there is a good reason for this, and we are seeing it now.

I'd love to see some stand alone products with this similar to what Audyssey had with their "pro kit". Same for Room Perfect for that matter. Processing power has always held back AVRs IMHO.