Active Room Treatment (ART) by Dirac

[slaweks]

I don’t buy this, lower in level i can go behind but so low to the point that it’s not audible?

If it’s not audible then the issues that it was counteracting weren’t audible either.

I think this depends on the number of support speakers and obviously how the algorithm itself works which Dirac has been almost suspiciously secretive about.

I guess they are inaudible, because we are talking about <=150Hz, so little directionality.

 

[Curvature]

I don’t buy this, lower in level i can go behind but so low to the point that it’s not audible?

If it’s not audible then the issues that it was counteracting weren’t audible either.

I think this depends on the number of support speakers and obviously how the algorithm itself works which Dirac has been almost suspiciously secretive about.

We know ART works from the measurements posted. Exactly how is proprietary and wider adoption will let us investigate. It's an old idea (impulse correction), but this is the first successful commercial implementation.

"Not audible" means not audible relative to direct sound. We know from reports that "not audible" is relative, since adjustment of support is a built-in capability. You can easily boost the antisound too high and cause audible issues.

 

[MickeyBoy]

I don’t buy this, lower in level i can go behind but so low to the point that it’s not audible?

If it’s not audible then the issues that it was counteracting weren’t audible either.

I think this depends on the number of support speakers and obviously how the algorithm itself works which Dirac has been almost suspiciously secretive about.

"If it's not audible then the issues it was counteracting weren't audible either." Unless I am missing something basic, this does not make sense. If something being cancelled out is audible before cancellation, then it and the cancelling signal should not be audible at all. The idea is that room modes should be cancelled out, not music signals. No?

 

[abdo123]

"If it's not audible then the issues it was counteracting weren't audible either." Unless I am missing something basic, this does not make sense. If something being cancelled out is audible before cancellation, then it and the cancelling signal should not be audible at all. The idea is that room modes should be cancelled out, not music signals. No?

Lets say Speaker 3 is sending out signals to correct Speaker 1.

You turn off Speaker 1, but still can’t hear the “correction” Speaker 3 is throwing into the room (because it assumes Speaker 1 is playing).

In this situation, do you think whatever speaker 3 is doing is impactful / audible if you can’t even “hear” the correction signals in a vacuum?

 

[abdo123]

"Not audible" means not audible relative to direct sound.

Well then this statement has absolutely zero merit to draw the conclusion from that the “correction” is very low in amplitude.

 

[dlaloum]

I don’t buy this, lower in level i can go behind but so low to the point that it’s not audible?

If it’s not audible then the issues that it was counteracting weren’t audible either.

I think this depends on the number of support speakers and obviously how the algorithm itself works which Dirac has been almost suspiciously secretive about.

If the correcting signal is audible, then it will utterly fail as an audiophile tool.... by definition, it should do its correction "inaudibly" (other than the desired corrective effects) - if you can hear the corrective signal, it is a failure.

 

[Emlin]

Lets say Speaker 3 is sending out signals to correct Speaker 1.

You turn off Speaker 1, but still can’t hear the “correction” Speaker 3 is throwing into the room (because it assumes Speaker 1 is playing).

In this situation, do you think whatever speaker 3 is doing is impactful / audible if you can’t even “hear” the correction signals in a vacuum?

In a vacuum you won't be able to hear yourself scream.

 

[GXAlan]

I'm now using a Monoprice HTP1... Which is also likely busted as biased/auto/etc switching on the inputs doesn't resolve the intermittent silence issue from an Nvidia Shield or Nintendo Switch.

I have a shield pro, but haven’t used it with my HTP-1 since going with an Apple TV. I do have a Nintendo Switch connected to my HTP-1 and have never had troubles.

With my AppleTV things were going great for the longest time and then I had trouble with it dropping to 1080p and then finally the ATV said that the HDMI transmission was unstable. Swapping cables fixed everything.

As much as we say digital is digital, I do think the HTP-1 is sensitive and even though it doesn’t support 8K, I find that using 8K certified cables fixed the ATV issue. I just use Monoprice generic 8K cables, so they are not expensive. That might fix your audio drop outs.

Separately, I had a Panasonic UHD player going into a CXA5100. Dropouts in audio with *some* UHD discs until I swapped the cables and then it was fixed too.

 

[Dj7675]

An observation from users of StormAudio's ART deployment - the support speakers, when properly configured, are not audible - that implies that the support signal going through them is at a substantially lower level (otherwise it would be audible! - which would spoil the effect!)

So it is pretty obvious.

During some of the Dirac interviews, the question was asked and answered, that the support signal was lower level and therefore not audible.

But more details and explanations would be nice!

This is true... support/cancellation signals are very low. An interesting thing to do is to play stereo content without upmixing. Support speakers are all still active and only playing cancellation signals. And one of the biggest things to remember is you have complete control over the level of the cancellation signals. For example if you want to configure the support speakers to play a lower level of cancellation signals, you can do that. For example my surrounds are a bit close to me. To make sure they didn't become audible I lowered the level on just the surrounds and left the other speakers at the default level.

 

[stevenswall]

I have a shield pro, but haven’t used it with my HTP-1 since going with an Apple TV. I do have a Nintendo Switch connected to my HTP-1 and have never had troubles.

Try the Gameboy switch apps with their mono midi sounds. Causes issues. On the shield it's watching Shogun or Invincible.... Or really anything.

I'm using 8k cables.

Hopefully there is something stable with Dirac ART soon as Storm Audio prices and up are past what I'm willing to put up with for those kinds of products. Maybe in another decade or so I'll be able to afford something like that and hope they don't have issues.

 

[GXAlan]

Try the Gameboy switch apps with their mono midi sounds. Causes issues. On the shield it's watching Shogun or Invincible.... Or really anything.

Interesting. I admittedly have not tried any of the classic games.

Will try my NVIDIA Shield as well out of curiosity. I am a big NVIDIA fan from the NV1 era, but the lack of ads on the app home Screen for AppleTV is worth the price of admission. I have toddlers and it’s annoying when the AndroidTV ads show horror films or things that wouldn’t be appropriate for kids. Even when advertising MA/R rated shows or movies on Apple, they seem to do a better job picking an innocuous poster.

I'm using 8k cables.

+1. I love the sound and feature set of my HTP-1 but it does require the occasional reboot which none of my mainstream Av gear requires. From what I have read, StormAudio and Trinnov are reliable as well. If you don’t have issues when setting it up, the unit will perform as configured indefinitely. Trinnov did have some sort of issue with Klipsch speakers, maybe @Buckeye Amps setup if I remember correctly.

Hopefully there is something stable with Dirac ART soon as Storm Audio prices and up are past what I'm willing to put up with for those kinds of products.

+100

@TimoJ is one of the rare alpha testers who is also publicly discussing HTP-1/ART. It sounds like there are just interface issues more than anything completely deal breaking.

I bought DLBC last November, wasn’t super happy and asked for a refund. @Flak asked for my project file and at my request ran it through ART to see what would happen if/when ART is available for the HTP-1 and the results looked really promising.

Yet, even the Focal Astral 16 is lacking Dirac ART and all of the complaints going from Dirac 3.4 to the current version seem to be related to modifications to the software related to broader ART support.

Based upon the Focal Astral being slow to have ART and no other company offering ART, I bet the problem is something on Dirac’s end whether technical or commercial/financial (in terms of what they ask licensees to pay to offer ART).

It’s a real race to see who is “next” after StormAudio! I would also like to see how ART works with more ordinary rooms and speakers…

 

[Buckeye Amps]

+1. I love the sound and feature set of my HTP-1 but it does require the occasional reboot which none of my mainstream Av gear requires. From what I have read, StormAudio and Trinnov are reliable as well. If you don’t have issues when setting it up, the unit will perform as configured indefinitely. Trinnov did have some sort of issue with Klipsch speakers, maybe @Buckeye Amps setup if I remember correctly.

There was a weird issue with the Trinnov paired with NC502MP modules and Klipsch 180LCR In ceiling speakers. Only with that combination. Never found out why, as I switched to Storm Audio and it has been fantastic.

 

[Davide]

The discussion on the audibility of the correction waves is interesting. Technically the low frequencies have a long length so the area of effectiveness remains large and the sound pressure differential (with mains) minimum.
It is more difficult to understand how it remains sufficiently inaudible as the frequency rises, since Dirac says it could work up to 4khz...
Another interesting thing is to understand how at low frequencies the distortion of a small support speaker (like 6.5") interacts with the less distorted sound of the sub(s).

 

[dlaloum]

The discussion on the audibility of the correction waves is interesting. Technically the low frequencies have a long length so the area of effectiveness remains large and the sound pressure differential (with mains) minimum.
It is more difficult to understand how it remains sufficiently inaudible as the frequency rises, since Dirac says it could work up to 4khz...
Another interesting thing is to understand how at low frequencies the distortion of a small support speaker (like 6.5") interacts with the less distorted sound of the sub(s).

Yes, I think low distortion bass is going to be more important - although distortion in the bass has a relatively low level of audibility (ie it has to get to a high % before it becomes audible) - its impact on cancellation might be greater - and it might become audible.

On the other hand the SPL's neeeded for support are lower, and typically distortion rises/falls with SPL.

Another long term observation is that woofers/subs with smaller drivers and sealed cabinets tend to have lower distortion (yes I know, a massive generalisation) - but also lower SPL's - which might suit support speakers very well indeed.

 

[Curvature]

The discussion on the audibility of the correction waves is interesting. Technically the low frequencies have a long length so the area of effectiveness remains large and the sound pressure differential (with mains) minimum.
It is more difficult to understand how it remains sufficiently inaudible as the frequency rises, since Dirac says it could work up to 4khz...
Another interesting thing is to understand how at low frequencies the distortion of a small support speaker (like 6.5") interacts with the less distorted sound of the sub(s).

The correction is most effective for the point in space where the measurement happens. As you get farther away, and as the frequency in question increases, the correction will lose effectiveness. It works the same way with normal EQ. At LF modal behavior will determine how consistent the frequency response, and corrections, will be for a given area.

The possibility of correction up to 4kHz is relevant only for cars (at least right now). In a car, it is known with a reasonable degree of accuracy where the listeners head and ears will be located. Not so in a room.

The antisound coming from support speakers is just sound, but lower in level and delayed. Regular acoustics and psychoacoustics apply. Masking will be effective unless, because of a listener's or speaker's position, the antisound is stronger than expected. There will probably have to be a compromise between the main listening position and others, especially in living rooms.

If distortion is a problem, you can adjust LF extension of the support speakers.

In general ART seems like it needs more listening and fiddling with settings than regular EQ, even for the experienced.

 

[jtatknox]

A simplistic explanation for the audibility question is that 1) you won't hear the support signals if they are calibrated because they are creating destructive interference (both main and support signals are attenuated) and 2) attenuating room modes focuses on the problematic aggregation of low-frequency energy, which is prone to reinforcement within bounded spaces. Such modal resonances lead to disproportionate amplification of certain low-frequency bands, distorting the acoustic fidelity of the space. The strategic deployment of support loudspeakers aims to introduce inverse phase sound waves at targeted low-frequency modes. This intervention seeks to precipitate an early interference with the modal energy, effectively curtailing the Q-factor (quality factor) of these resonances. Consequently, this technique is instrumental in diminishing the reverberation time associated with low-frequency sound waves, thereby achieving a reduction in their persistence and perceptual dominance within the acoustic environment.

 

[pogo]

ART assumes a compression-free playing system at any volume during co-optimisation. If this is not guaranteed, audible correction artefacts may even be produced!

 

[dlaloum]

ART assumes a compression-free playing system at any volume during co-optimisation. If this is not guaranteed, audible correction artefacts may even be produced!

There is no such thing as a compression free playing system

It is always a question of how much or how little compression.

It's like talking about a frictionless surface... a pragmatic impossibility

So the question is, what level of imperfection will ART tolerate and still work properly?
And what levels of imperfection can be tolerated while achieving close to optimum performance?

 

[Frank207be]

There is no such thing as a compression free playing system

It is always a question of how much or how little compression.

It's like talking about a frictionless surface... a pragmatic impossibility

So the question is, what level of imperfection will ART tolerate and still work properly?
And what levels of imperfection can be tolerated while achieving close to optimum performance?

Using enough speakers that can play clean up to 110-120dB and as low as 25-30Hz?

 

[pogo]

Don't forget the amps/AVR amp sections