Hmm I was running 3.13.16... must update...
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Active Room Treatment (ART) by Dirac
- Thread starter Flak
- Start date
Just downloaded "dirac-latest" and installed it, which puts me on 3.13.16....
I wonder where 3.14 is?
EDIT: - OK found it, it is the latest BETA test version, not yet production... gonna give it a try see if that changes the display
OK now running 3.14 - loaded my ART project, went to filter design - and the filter # is right where I expected it to be....
Maybe needs a message to Dirac support?
I don't know, but ChatGPT seems to know, and 240 is the number:
When the number of active channels is 16, the calculation is straightforward.
Dirac ART uses the following principle:
cross terms = N × (N − 1)
where N is the number of active channels.
Calculation
- N = 16
- 16 × 15 = 240 cross terms
Interpretation
- 240 is the theoretical and practically required number for all channels to be able to support all other channels in the ART algorithm.
- If Dirac ART reports a lower number, it means:
- a hardware or DSP limitation, or
- optimization performed by Dirac where not all cross terms are enabled.
Thank you for verifying. I will contact Dirac Support.
Not going to defend D&M for limited filter implementation - more is definitively better. From practical perspective though, measurements on D&M gear look quite good.
@TimoJ - do your subs fire a bit ahead of the mains as that happens with some D&M systems?
How do I check that?
There are couple of ways in REW. I usually use this diagram as can see other things as well.
Dotted line on the bottom shows that most of the spectrum is delayed a bit compared to subs. Disregard 20-30hz bump at the beginning of sub response. Not good but it is what it is. Looks like after ART handover to Dirac Live, it gets a bit bumpy.
Dotted line on the bottom shows that most of the spectrum is delayed a bit compared to subs. Disregard 20-30hz bump at the beginning of sub response. Not good but it is what it is. Looks like after ART handover to Dirac Live, it gets a bit bumpy.
Thanks Timo. Looks like it exhibits the same behaviour, so likely Dirac matter and unrelated to implementation. I run Dirac full range, thus I have a delay all the way up to 20khz.
Not sure how relevant this point is, but it's out there. Perhaps something that Dirac could keep on trying to do - better time alignment between ART and DL. Sure they tried already, but still could use some work.
EDIT: My graph was 4 subs and 7 out of 9 beds, but your LF is probably indicative enough to highlight the issue. Not trying to blow it up or say it is audible, but I do like to know how it measures. Ultimately I decide based on the ear, between graphs that look acceptable, albeit different.
Not sure how relevant this point is, but it's out there. Perhaps something that Dirac could keep on trying to do - better time alignment between ART and DL. Sure they tried already, but still could use some work.
EDIT: My graph was 4 subs and 7 out of 9 beds, but your LF is probably indicative enough to highlight the issue. Not trying to blow it up or say it is audible, but I do like to know how it measures. Ultimately I decide based on the ear, between graphs that look acceptable, albeit different.
It doesn't reflect what happened in my setup...
Initially configured as 5.1.4 - which using your formula should provide 10 active channels and 90 cross terms - but the default (everything supports everything else) configuration reported 98 in use
Dropping off the heights, making it 6 or 7 channels (depending on # of subs) - the calculation returns 30 or 42 cross terms - but actual implementation had over 50.
So yeah, that ain't it... more information needed!
Is this for your front left? (Edited) Never mind. I saw you edited your post.
Last edited:
I don't know Dirac's specific terminology, like what they mean by cross terms or even by filter honestly, but just in terms of signal processing if we define a "filter" as the tool that modifies a given channel's signal to be played out of a single speaker a fully supported system of N channels would require N*N filters. N filters for the speakers that are that channel themselves (traditional room correction), and N-1 filters for each additional speaker being used as a support. So a 5.1.4 system would require 100 filters to be fully supported.
Unfortunately that means in a 15.4 setup my previous calculation was optimistic. It's not 15*4 (4 subs supporting 15 channels) + 3 (3 subs supporting sub 1) = 63 filters, it's 15*4 + 3 + 15 (channels) + 1 (sub 1) = 79 filters.
But if you saw that 6 or 7 channels both required more than 50 filters my N*N is still optimistic to how they are actually counting things.
jacktracks707
New Member
- Joined
- Oct 3, 2024
- Messages
- 4
- Likes
- 11
Ran my first Dirac Art calibration on my smaller 5.2 system last night. Wanted to play with the software and dial in this room before doing my main 9.4.4 system. Gear in this room is Denon 3800. Kef Q950 fronts are group 1. Listed response is 44hz. Group 2 is Kef R2 Meta. Listed response is 67hz. Group 3 is surrounds, Kef Q1 Meta. listed response is 51hz. I have two Klipsch RP1400s catty corner in front and back of room. Umik1. I have REW but have never ran any sweeps personally. (calibrators have). I could probably figure that out tomorrow as I know it's helpful.
My subs go down to 16hz in this room and 13hz and the other room. When checking the infrasonic bypass button where exactly do you do that, and to which groups. So far I have only enabled that checkbox while I am on my subwoofer group and for the subwoofer only. Should I also be going into the each of the other groups and enabling that checkbox for the subwoofers as well?
Should I apply target curves to all groups? So far 6db, 8db, 10 db harman curve is sounding the best. Is that too much to apply to all groups?
Should I adjust my FSL closer to the listed spec or keep them conservative like below?
Current settings are as follows. Would love some tips on a better place to start or if anything seems really out of whack. First take is bass got nuked.
Group 1 is supporting itself at -6db, fsl = 65hz. Group 1 is also supported by group 3 at -6db and FSL = 70hz, and Group 4 at -18db, FSL=20hz.
Group 2 is supported by Group 1 at -6db, fsl=65hz and supported by group 4 at -18db, fsl=20hz
Group 3 is supported by group 1 at -6db, fsl- 65hz, supporting itself at -12db, fsl=70hz, and by group 4 at -18db, fsl=20hz
Group 4 is supporting itself at -24db, fsl=20hz, it is also being supported by group 1 at -5db, fsl = 65hz.
Bass was lacking with anything less than the 8db hamrmn curve applied. I plan on re doing my measurements tomorrow with better volume/levels than I did on my first shot. I will follow the guide to a tee this time. Center channel also seems a little hot/harsh. Should I limit the frequency on this group with the curtain? What frequency?
Thanks for any help. I'm happy to post screenshots and take some REW measurements just don't exactly know what Im doing.
My subs go down to 16hz in this room and 13hz and the other room. When checking the infrasonic bypass button where exactly do you do that, and to which groups. So far I have only enabled that checkbox while I am on my subwoofer group and for the subwoofer only. Should I also be going into the each of the other groups and enabling that checkbox for the subwoofers as well?
Should I apply target curves to all groups? So far 6db, 8db, 10 db harman curve is sounding the best. Is that too much to apply to all groups?
Should I adjust my FSL closer to the listed spec or keep them conservative like below?
Current settings are as follows. Would love some tips on a better place to start or if anything seems really out of whack. First take is bass got nuked.
Group 1 is supporting itself at -6db, fsl = 65hz. Group 1 is also supported by group 3 at -6db and FSL = 70hz, and Group 4 at -18db, FSL=20hz.
Group 2 is supported by Group 1 at -6db, fsl=65hz and supported by group 4 at -18db, fsl=20hz
Group 3 is supported by group 1 at -6db, fsl- 65hz, supporting itself at -12db, fsl=70hz, and by group 4 at -18db, fsl=20hz
Group 4 is supporting itself at -24db, fsl=20hz, it is also being supported by group 1 at -5db, fsl = 65hz.
Bass was lacking with anything less than the 8db hamrmn curve applied. I plan on re doing my measurements tomorrow with better volume/levels than I did on my first shot. I will follow the guide to a tee this time. Center channel also seems a little hot/harsh. Should I limit the frequency on this group with the curtain? What frequency?
Thanks for any help. I'm happy to post screenshots and take some REW measurements just don't exactly know what Im doing.
I checked the Dirac's internal value with those setups and it was 90, 30 and 42. So the formula seems to be correct. But something else is added on top of that when it's displayed.
What exact number did you get with 5.1.0 and 5.2.0 setups? Knowing that would maybe allow to see if the added number has some logic.
5.2.4 with subs in a single group (as per default - splitting them into seperate groups used more filters!) - with the heights being supported but not supporting anything... 63 /98
Note that this is not a true 5.2.0 config, as the heights are present and being supported by subs and base layer - I merely removed them from supporting all other groups (which dropped it from 98/98 down to 63/98... and provided far far better results!)
Can't open my 5.1.4 projects... I would need to reconfigure the AVR to 5.1.4 before Dirac will load the project! - from memory it was 53 (I think)
The number I can check is the max number, it won't change when support or groups are changed.
So, in order to (maybe) figure this out, you would need to check what number a real 5.2.0 or 5.1.0 setup shows with full support enabled.
That may have to wait a while....
I've posted this in the main Dirac live forum but no one can explain what I'm seeing
I'm running 9.4.4 in a dedicated room with acoustic treatment including bass traps in the rear corners
The subs are in the corners.
The longitudinal room mode is around 21.4Hz
When I measure the results with REW, both DLBC and ART show higher distortion and worse decay than no Dirac around the region of the room mode.
I attach the .mdat file for those interested
The most telling result looks to me to be taking the Decay measurement at 200ms
No EQ has a decay of 25db aT 21.4Hz Distortion 0.22%
ART with flat curve (0db rise) gives only 5db Distortion 3.23%
ART with 10db Harman curve gives only 2db Distortion 2.26%
Results are slightly better with DLBC but still much worse than no Dirac
I've run multiple calibrations across MLP, front row, both rows and the basic principle doesn't change.
I thought maybe Dirac filters were the cause but 10Db Harman maxes out filters at 10db andFlat uses no gain filters at all and they both have poor decay/increased distortion
If any one has any ideas?
drive.google.com
I'm running 9.4.4 in a dedicated room with acoustic treatment including bass traps in the rear corners
The subs are in the corners.
The longitudinal room mode is around 21.4Hz
When I measure the results with REW, both DLBC and ART show higher distortion and worse decay than no Dirac around the region of the room mode.
I attach the .mdat file for those interested
The most telling result looks to me to be taking the Decay measurement at 200ms
No EQ has a decay of 25db aT 21.4Hz Distortion 0.22%
ART with flat curve (0db rise) gives only 5db Distortion 3.23%
ART with 10db Harman curve gives only 2db Distortion 2.26%
Results are slightly better with DLBC but still much worse than no Dirac
I've run multiple calibrations across MLP, front row, both rows and the basic principle doesn't change.
I thought maybe Dirac filters were the cause but 10Db Harman maxes out filters at 10db andFlat uses no gain filters at all and they both have poor decay/increased distortion
If any one has any ideas?
CAL 36 TESTS.mdat
drive.google.com
As a starting point it might be worth while measuring your subs distortion at the reference level...I've posted this in the main Dirac live forum but no one can explain what I'm seeing
I'm running 9.4.4 in a dedicated room with acoustic treatment including bass traps in the rear corners
The subs are in the corners.
The longitudinal room mode is around 21.4Hz
When I measure the results with REW, both DLBC and ART show higher distortion and worse decay than no Dirac around the region of the room mode.
I attach the .mdat file for those interested
The most telling result looks to me to be taking the Decay measurement at 200ms
No EQ has a decay of 25db aT 21.4Hz Distortion 0.22%
ART with flat curve (0db rise) gives only 5db Distortion 3.23%
ART with 10db Harman curve gives only 2db Distortion 2.26%
Results are slightly better with DLBC but still much worse than no Dirac
I've run multiple calibrations across MLP, front row, both rows and the basic principle doesn't change.
I thought maybe Dirac filters were the cause but 10Db Harman maxes out filters at 10db andFlat uses no gain filters at all and they both have poor decay/increased distortion
If any one has any ideas?
CAL 36 TESTS.mdat
You may find that you need to constrain the frequency response a little more than expected to get it operating within a low distortion envelope
With sealed subs you can probably get closer to its spec frequency than with ported ones...
My theory is that if the support speakers distortion gets too high, then the results of the ART sum will be worse... as the outcome is not the calculated value, but some distorted value thereof - and that will result in distortion on top of the distortion of the speaker itself.
Hence my Surround Rears - a ported speaker specced at 35Hz, is limited to 40Hz - to keep within a lower distortion envelope
My mains are specced at 24Hz - sealed speaker - limited to 30Hz
You can see the slight messiness just below 30Hz.... the subs are generating too much THD to be able to control it thoroughly (it's still very good though!)
Subs specced at 24Hz sealed - limited to 26Hz (I probably should limit them to 30Hz as the THD at 26 is 12% whereas at 30Hz it is down at 2.7%) with ART active the THD on the LFE drops to 6.6% at 26Hz
Here is the SUB / LFE
I do lose some low end, and the decay control is a bit patchy below 30Hz but from 30Hz up it is pretty clean.
I found that the measured performance for my full range mains and surrounds at their bottom end, THD is lower than when measured standalone (without ART support) - probably because they are getting support from other speakers, so each individual speaker is less stressed (running at lower SPL, and therefore lower distortion)
Similar threads
