Dirac in small room. Should I expect more?

[antcollinet]

So I've finally got round to re-doing the dirac measurements and filtering after converting my small study into a mini cinema.

The room is about 2.8 by 3.5 meters (definately non ideal. Two side walls are glass (yes, it gets worse). Left hand wall has curtains. Right hand wall is a door into the living room, so does not. Floor is carpted - a mini sofa is in the listening postition, with shelving behind it offering some diffusion I guess. There is also a small window on the back wall with blind over it. Front wall has a large TV with the main speakers (DALI oberon 5) on either side. A single 12 inch sub (BKElec XXLS400-DF) completes the arrangement on the left hand side, in front of the left speaker. With the room/tv and sofa sizes there is no real scope for position adjustment of anything in this small room.

Amp is an aging Sony AVR, and I've used the AVR auto cal to integrate the Sub (Digital Cinema Auto Calibration??)

I've then run Mini DSP Flex Dirac on top of that treating the speakers/sub as an integrated speaker set.



In the following 4 screenshots, I show the DIRAC filter design page, then meausurements from REW of:
1 - Dirac OFF and AVR set to analoge direct (no processing)
2 - Dirac Off but AVR processing on (the result of auto cal)
3 - Dirac and Auto cal both enabled.

1/12 smoothing is applied.

To be honest the sound is significantly improved with DIRAC on - even compared to the auto cal. But the plots are significantly less smooth than others I've seen - and there is that big dip at 70Hz or so

Are there any obvious improvements I could get? Thanks for looking.

 

[ozzy9832001]

Despite what some will tell you, a true null can't be boosted out of. It's based on your position in the room and the fact that the sound wave fits exactly inside it. You could technically absorb or cancel it out with opposing energy (this is difficult and still creates moderate ringing). Dirac knows this and will not apply boosts to a null if the loss is greater than like -10dB. Doing so, will cause a crazy amount of ringing and combine with the higher harmonics making a potential mess of the sound field.

The room dictates the response. I believe dirac is configured to really work with the curve you have, so it seems to center the SPL without changing the peaks and valleys.
When I tried DIRAC, I didn't like the default curves, so I like it handle anything higher than 350hz and I manually configured the rest. It's a long process, but I think it's worth it. My room is small so I have a ton of room modes that start stacking up in the mid/high bass and it can get really muddy when listening to music. Most times this won't show with measurements alone.

You also have to make sure DIRAC is synced with the microphone properly. I had to lower the master gain to 17.5 dB with mic at 100% gain to match what REW and a separate dB meter say. DIRAC was +8dB which lead to some interesting clipping issues.

I'm hoping you like bass since it's really quite loud compared to the rest of the sound.

 

[antcollinet]

Despite what some will tell you, a true null can't be boosted out of. It's based on your position in the room and the fact that the sound wave fits exactly inside it. You could technically absorb or cancel it out with opposing energy (this is difficult and still creates moderate ringing). Dirac knows this and will not apply boosts to a null if the loss is greater than like -10dB. Doing so, will cause a crazy amount of ringing and combine with the higher harmonics making a potential mess of the sound field.

The room dictates the response. I believe dirac is configured to really work with the curve you have, so it seems to center the SPL without changing the peaks and valleys.
When I tried DIRAC, I didn't like the default curves, so I like it handle anything higher than 350hz and I manually configured the rest. It's a long process, but I think it's worth it. My room is small so I have a ton of room modes that start stacking up in the mid/high bass and it can get really muddy when listening to music. Most times this won't show with measurements alone.

You also have to make sure DIRAC is synced with the microphone properly. I had to lower the master gain to 17.5 dB with mic at 100% gain to match what REW and a separate dB meter say. DIRAC was +8dB which lead to some interesting clipping issues.

I'm hoping you like bass since it's really quite loud compared to the rest of the sound.

Well my initial stab had the bass shelf set at +7dB, I could bring that down a notch or two.


EDIT : Just dropped that down to +3dB and it does sound a little cleaner. I'll try an REW measurement tomorrow on that.

 

[ReDFoX]

I have a 2.2x5.5 room with a pair of KH80 + DDRC24.
This was the most "flat" response I got from Dirac visualiser, even without any significant treatment.

I strongly suggest you to do MMM measurement for each channel, make filters in REW (mainly in midbass/bass region) and then upload them to miniDSP. This way you'd help Dirac, since it dont have infinite averages unlike you have and I feel like Dirac still misses sometimes in bass region in terms of determining correct filter frequency
Also you can manually allign LR channels with REW and white noise. Dirac time allignment was off in my room in like 70% of the calibrations I did

 

[antcollinet]

EDIT : Just dropped that down to +3dB and it does sound a little cleaner. I'll try an REW measurement tomorrow on that.

And I've just discovered that I had a Graphic EQ set with a rock preset. Annoying since I'll now have to repeat the measurements with that disabled.

 

[mglobe]

And I've just discovered that I had a Graphic EQ set with a rock preset. Annoying since I'll now have to repeat the measurements with that disabled.

Welcome to my world.

 

[ZolaIII]

This is two times run REW PEQ (in JRiver manually inputed) with QA Q3030i and one 10" sub in similar sized room (and also problematic).

https://www.audiosciencereview.com/forum/index.php?attachments/sad-jpg.290947/

Advice is first do a REW PEQ into MiniDSP Flex (if it can do Dirac and PEQ simultaneously) squishing bass peak room mods primary (47 & 90 Hz) and then run Dirac Live over it or export FIR filter to it (even better but a little more complicated). What's the crossover order and frequency? Asking because you might try putting a relatively high one (100 Hz) in order to try and see if it will feel that 70 Hz hole a bit.

 

[Deleted member 48726]

Hi @antcollinet

Another experiment you could do with Dirac is force cutting.. Like this

 

[neRok]

3.5m = 49Hz first room mode. You see the multiples 2x and 3x are extremely strong in this plot too. That's because you are sitting so close to the back wall that all 3 modes are near full power, like the following graphic shows. If you can move forward slightly, you should be able to bring ~150Hz down nicely,~100Hz down a fair bit, but unfortunately you won't improve ~50Hz much.