Active Room Treatment (ART) by Dirac

[markus]

My system looks like this:
View attachment 530882

Same behavior. Interesting.

 

[janbth]

@RenPa What is "TC"?

 

[EWL5]

TC = Target Curve

 

[RenPa]

This is a comparison of a step response of that ART measurement (red) against a simulated "ideal" loudspeaker (green).

So DLBC doesn't calculate the filter like Art which is expected. The sound is so much different! Like I said before, since Art I've lost my points of reference when it comes to listening; everything is very new. I don't dare say it's better (which it actually is to be fair)—for now, I'll just say it's different, extremely different!

 

[mmares056]

The subwoofer is firing before the main impulse, about 15ms early. So it's not time aligned. I have seen enough ART measurements to suspect this is a deliberate choice by Dirac to do that.

cc: @markus

Okay so knowing that the verified impulse response measurements confirm the subs are firing early before the main impulse and as you stated that you have seen enough ART measurements to suspect this is a deliberate choice by Dirac to do that do we have a consensus to believe that in order for ART accomplish a flat phase optimization it needs to make the sub(s) super early? Am I correct that the fist rise in the impulse response of the sub(s) should be aligned at t=0 since that is when the energy from the sub(s) start to arrive at our microphone or ears?

We can agree that ART's powerful logarithm seeks to achieve an ideal IR that shows the earliest and strongest main peak, followed by rapid settling toward zero and the Step Response confirms that by showing a fast initial rise, minimal ripple/oscillation, and no late bumps or spikes. We can also confirm by looking at the Spectogram Wavelet when a system is well time aligned. The objective data (Screenshots) shows that is what's happening in the time domain for speakers but for sub(s) it is different as ART makes the sub(s) start early to accomplish the flat phase for optimal sub(s) to speakers integration and whatever weird stuff ART is doing in this region seems to be working. Subjectively, the current 5ms LFE delay is not audible to my ears and the bass is well balanced and it sounds tight and natural in my room. I'm still wondering what ART is going to do to fix the 5ms LFE delay caused the 120Hz LPF. ART is intriguing and it is fun learning how room correction algos work. BTW the REW eBook is an excellent educational source. Thank you.

ART Subs + Center IR

ART Subs + Center Head of the Step Response

ART Subs + Center Tail of the Step Response

ART Center Spectogram Wavelet

ART Center Excess Group Delay

 

[markus]

I'm still wondering what ART is going to do to fix the 5ms LFE delay caused the 120Hz LPF.

Maybe make the LPF optional

 

[Keith_W]

Okay so knowing that the verified impulse response measurements confirm the subs are firing early before the main impulse and as you stated that you have seen enough ART measurements to suspect this is a deliberate choice by Dirac to do that do we have a consensus to believe that in order for ART accomplish a flat phase optimization it needs to make the sub(s) super early? Am I correct that the fist rise in the impulse response of the sub(s) should be aligned at t=0 since that is when the energy from the sub(s) start to arrive at our microphone or ears?

I do not have expertise in MIMO. I can vaguely tell you how it works, but I do not have an understanding of the mathematical principles behind it. I was given some papers, I tried to read it, but I did not get very far. Smarter people than me have figured it out, and I am aware that very soon a MIMO competitor might come on the market. So perhaps if they see this thread they might want to reply.

As for where subs should be aligned at t=0, it should be at the start of the larger deflection. Be careful that the polarity of the sub isn't inverted because that can easily trick your eyes.

I'm glad you found the eBook helpful.

 

[EWL5]

Smarter people than me have figured it out, and I am aware that very soon a MIMO competitor might come on the market. So perhaps if they see this thread they might want to reply.

There's at least one and they posted in this thread:

Looking for Dirac ART and Trinnov users

Would anyone here be kind enough to help compare the performance of my MIMO speaker calibration software to Dirac ART or Trinnov? Mine is fully automatic and also takes your distortion measurements into consideration as well. I know it probably seems a little bit outrageous and silly to think...

www.audiosciencereview.com

 

[RenPa]

I have searched on the net and got pros and cons about the following question. Is it a good practice to EQ my 2 SVS PRO PC2000 with internal eq app to reduce large peaks with large Q (< 2) before measuring with Dirac for Art filters? Does a consensus exist about equing or equing before Art?

 

[90sNintendoKid]

I have searched on the net and got pros and cons about the following question. Is it a good practice to EQ my 2 SVS PRO PC2000 with internal eq app to reduce large peaks with large Q (< 2) before measuring with Dirac for Art filters? Does a consensus exist about equing or equing before Art?

I would only do this if ART has difficulty reducing the peaks or hitting your target curve and I would verify with post EQ measurements

 

[RenPa]

I would only do this if ART has difficulty reducing the peaks or hitting your target curve and I would verify with post EQ measurements

I tried and it sounds different but nothing in REW show significant differences: spl, spectrogram, excess GD. Only distortion below 100Hz a little higher but still below 1% . Sub left yellow is equed in svs app

 

[ernestcarl]

ART Subs + Center Head of the Step Response

This misalignment of the LFE relative to the bass-managed mains might very well be just a measurement artifact/error. Although, SPL level seems a bit off too as I would expect there to be a 10dB gap between the LFE signal and bass-managed mains (~4dB difference is just not enough).

 

[janbth]

I have searched on the net and got pros and cons about the following question. Is it a good practice to EQ my 2 SVS PRO PC2000 with internal eq app to reduce large peaks with large Q (< 2) before measuring with Dirac for Art filters? Does a consensus exist about equing or equing before Art?

I have two SVS subs as well. Running ART, not EQing the subs with the SVS app currently, but I don’t see why not. In my previous setup with a NAD M33 amplifier, I EQd my SVS subs and ran regular Dirac Live on top of that. To compute the filters, I used a spatial average of a number of point measurements in REW. I applied the same filters on both subs, and this created a sort of global EQ. These were high Q filters, aimed to cut some nasty peaks in my room. I used the next layer of Dirac Live to apply EQ for each listening position - one preset for each of the listening positions. This worked quite well, and I don’t see why it shouldn’t work with Dirac ART too.

 

[janbth]

I have two SVS subs as well. Running ART, not EQing the subs with the SVS app currently, but I don’t see why not. In my previous setup with a NAD M33 amplifier, I EQd my SVS subs and ran regular Dirac Live on top of that. To compute the filters, I used a spatial average of a number of point measurements in REW. I applied the same filters on both subs, and this created a sort of global EQ. These were high Q filters, aimed to cut some nasty peaks in my room. I used the next layer of Dirac Live to apply EQ for each listening position - one preset for each of the listening positions. This worked quite well, and I don’t see why it shouldn’t work with Dirac ART too.

Maybe I was too fast with my post. I just remembered why I dropped EQing the subs when moving to Dirac ART. The problem is that the ART algorithm is proprietary, and we don't know if EQing subs upstream might even reduce the ability of ART to do its thing. Better be careful, and only try it if there is a specific problem. (This was already said by @90sNintendoKid!)

 

[markus]

Maybe I was too fast with my post. I just remembered why I dropped EQing the subs when moving to Dirac ART. The problem is that the ART algorithm is proprietary, and we don't know if EQing subs upstream might even reduce the ability of ART to do its thing. Better be careful, and only try it if there is a specific problem. (This was already said by @90sNintendoKid!)

If EQ is applied upstream of DL, DL won’t see it during calibration. During playback, the signal is still equalized by the upstream EQ, but this shouldn’t negatively affect ART.

 

[TimoJ]

Isn't that downstream if sub's own EQ is used? So DL will see it.

 

[TheKoC]

Depends on how you look at it I guess

Sub-EQ is done before ART-EQ, but once you have done ART, the Sub gets a signal changed by ART, and the original Sub-EQ (done without ART) isn't really valid anymore. I wouldn't do it without a very specific reason to do so (like limiting the anechoic sub-output), not for room-dependend EQ.

 

[markus]

Depends on how you look at it I guess

It only depends on how the signal flows. There’s no ambiguity here.
The only ambiguity comes from using “before” or “after,” which can refer either to time or physical order. “Upstream” and “downstream” don’t have that double meaning.

 

[90sNintendoKid]

Maybe I was too fast with my post. I just remembered why I dropped EQing the subs when moving to Dirac ART. The problem is that the ART algorithm is proprietary, and we don't know if EQing subs upstream might even reduce the ability of ART to do its thing. Better be careful, and only try it if there is a specific problem. (This was already said by @90sNintendoKid!)

I agree, this is a difficult one because with multiple subs, EQing individually before something like ART optimises delay and phase might mess things yes.

 

[janbth]

If EQ is applied upstream of DL, DL won’t see it during calibration. During playback, the signal is still equalized by the upstream EQ, but this shouldn’t negatively affect ART.

Isn't that downstream if sub's own EQ is used? So DL will see it.

Yes, I guess downstream is correct - sorry! The subs are seen by Dirac when making measurements, and taken into account when Dirac computes the filters.