Should we correct to Schroder, or full range?

[ernestcarl]

The measurement is with sub crossed at 80. I didn’t measure the sub correction output.

@Holdt

Assuming the Left speaker is high-passed at 80 Hz, there would be no apparent 10 dB boost in actuality:

80Hz HPF

*HOWEVER, if other EQ weren't disabled at all, then the +10dB boost is as is.

 

[jhwalker]

I have a semi opening to another room on one side. So not me either. I'm convinced I need the reflections to be able to correctly correct, so to speak.

Mine is WAY irregular LOL

It's basically part of an open plan living area, so only "closed" on one side (the side the speakers are closest to). One wall has two large (floor to ceiling) windows + a fireplace, and the other two walls are basically open: one to the kitchen / dining area and one to the main hallway that goes to the back of the house. In addition, there is both an irregular ceiling height (HVAC ducting retrofitted AFTER the house was built) AND a sloping ceiling - starting at about 8' near the windows up to 11' feet at the room boundary leading into the dining room.

 

[theREALdotnet]

And yet, using Dirac is not simple at all...

I was about to say! Ease of use is not exactly Dirac’s strong point. I don’t find the UI particularly intuitive and there are plenty of pitfalls awaiting the unwary. I‘d say it’s almost impossible to get a good result as a novice by just following the bouncing ball, so to speak. It starts with the volume calibration – plenty to do wrong there, and it took me a few articles (from Dirac and others) and Youtube tutorials to optimise that step alone.

For me, the value of Dirac (compared to taking the REW+rePhase or similar approaches) is the substantial reduction in the amount of work one has to do, and the (presumably) built-in intelligence about EQ, FIR vs PIR, what not to touch etc. For me, doing REW+rePhase each time I change my speaker or sub placement, adjust toe-in, change room furnishings etc. (all of which I do a fair bit), would be an excessive amount of effort. Dirac, the more I’m familiar with it and know how to correctly use it, streamlines this enormously.

Regarding Schroeder or not, for each measurement and cross-over point in Dirac I always generate three filters – one with a curtain at 250Hz, one at 500Hz and one full range. The differences between them are not subtle, but always confined to tonality. There are no weird phase effects or diminished soundstage and imaging with the full-range filter. I tend to select the filter that sounds best with the given program material.

With REW, on the other hand, I’ve been known to go overboard (before I learned to look at group delay etc). I got my measured frequency response ruler-flat, but the music sounded like crap. Dirac seems to shield me from that mistake.

 

[ernestcarl]

I swapped the rear speakers for my desk's multi-channel HTPC setup, and the ff. is a breakdown of the full-range EQ and bass management result for only the Rear Left channel:





A single-point measurement reference at the MLP -- contrasted with frequency dependent windowing 7 cycles i.e. 'direct sound' -- is supplemented with fairly wide moving microphone measurements (larger than 1x1x0.5 meter LxWxH rectangular area) around the MLP.

The speaker is pointed quite a bit off-axis towards the opposing side-wall. Room has some acoustic treatment and is relatively on the dry side compared to many other rooms. I kept boosting to a minimum and applied a considerable cut in the sub's bass response -- pay attention to the reflection-decay with(out) EQ. Distance to Rear speaker is almost 3m and distance to sub is about ~1.8 meters.

-----

Below is rather an extreme case study where it may not be easily possible to separate completely "room EQ" and "speaker EQ". Speaker is the Center channel of the same 7.1c desk setup but sitting right in front and above the desk LED monitor ~0.7m distance from the mic/MLP. Speaker is a second-hand bought, and somewhat out-of-spec Fulcrum Acoustic RX699 PA unit absent the grill.




The room:

*I can't find the more detailed speaker positioning diagram I saved prior so this will have to do.

MDAT and rePhase project files: https://drive.google.com/file/d/1mslGGiELCupet7zJiH_TvFiQ5eirrrqD/view?usp=sharing

 

[OCA]

Of course the correction window can be limited, but it looks like it's doing mostly eq above 300Hz. Here's the MDAT if interested.
View attachment 347317

This is a speaker response (I am hoping with no DSP!) vs receiver response measurement. The first measurement would be a perfect sound card calibration file for REW

 

[Deleted member 48726]

@Holdt

Assuming the Left speaker is high-passed at 80 Hz, there would be no apparent 10 dB boost in actuality:

80Hz HPF
View attachment 347374

*HOWEVER, if other EQ weren't disabled at all, then the +10dB boost is as is.

Didn't see that there was a high pass involved.

 

[OCA]

MDAT and rePhase project files: https://drive.google.com/file/d/1mslGGiELCupet7zJiH_TvFiQ5eirrrqD/view?usp=sharing

That's one hell of a subwoofer going down almost flat below 10Hz!!!

Btw, I think it needs to kick in some 6.5ms earlier relative to your center speaker according to the analysis below.

1/1 smoothed group delay comparison points to about 6.5ms delay in the sub response:

and impulse graph confirms that it should be synced to the next peak:

 

[ernestcarl]

That's one hell of a subwoofer going down almost flat below 10Hz!!!

Btw, I think it needs to kick in some 6.5ms earlier relative to your center speaker according to the analysis below.

1/1 smoothed group delay comparison points to about 6.5ms delay in the sub response:
View attachment 347476

and impulse graph confirms that it should be synced to the next peak:

View attachment 347478

With many prior experiments I've been able to nearly flatten the response. It does come with the risk of getting to the point where artifacts can sometimes be heard. So it's not always a win-win situation.



Oh, yeah, and I have a long running obsession in trying to reduce the total system processing latency with FIR use. This time, I was able to squeeze it down from ~47ms previously to ~40ms. This is a priority because I frequently stream videos realtime e.g. youtube, netflix on the desk PC.

 

[Obizzz]

I was about to say! Ease of use is not exactly Dirac’s strong point. I don’t find the UI particularly intuitive and there are plenty of pitfalls awaiting the unwary. I‘d say it’s almost impossible to get a good result as a novice by just following the bouncing ball, so to speak. It starts with the volume calibration – plenty to do wrong there, and it took me a few articles (from Dirac and others) and Youtube tutorials to optimise that step alone.

For me, the value of Dirac (compared to taking the REW+rePhase or similar approaches) is the substantial reduction in the amount of work one has to do, and the (presumably) built-in intelligence about EQ, FIR vs PIR, what not to touch etc. For me, doing REW+rePhase each time I change my speaker or sub placement, adjust toe-in, change room furnishings etc. (all of which I do a fair bit), would be an excessive amount of effort. Dirac, the more I’m familiar with it and know how to correctly use it, streamlines this enormously.

Regarding Schroeder or not, for each measurement and cross-over point in Dirac I always generate three filters – one with a curtain at 250Hz, one at 500Hz and one full range. The differences between them are not subtle, but always confined to tonality. There are no weird phase effects or diminished soundstage and imaging with the full-range filter. I tend to select the filter that sounds best with the given program material.

With REW, on the other hand, I’ve been known to go overboard (before I learned to look at group delay etc). I got my measured frequency response ruler-flat, but the music sounded like crap. Dirac seems to shield me from that mistake.

I played around with the Dirac for Windows free trial earlier this week and the volume calibration step really could have used more pointers on how and what to adjust. Going to be a bit more careful there on my next attempt and make sure to read up on it.

With that said my initial attempt was good enough to convince me I want to use Dirac in my system so I ordered a MiniDSP Flex today so I can apply it to all my sources.

Even if I could get better results from doing everything manually I’m not really interested in putting in the time and effort right now. With my limited knowledge I’m more likely to screw things up than Dirac’s algorithms. A couple of filters to pull down the most obvious peaks in the lower bass area makes a massive difference though but Dirac gave me even better results with even less effort.

 

[Snarfie]

I was about to say! Ease of use is not exactly Dirac’s strong point. I don’t find the UI particularly intuitive and there are plenty of pitfalls awaiting the unwary. I‘d say it’s almost impossible to get a good result as a novice by just following the bouncing ball, so to speak. It starts with the volume calibration – plenty to do wrong there, and it took me a few articles (from Dirac and others) and Youtube tutorials to optimise that step alone.

For me, the value of Dirac (compared to taking the REW+rePhase or similar approaches) is the substantial reduction in the amount of work one has to do, and the (presumably) built-in intelligence about EQ, FIR vs PIR, what not to touch etc. For me, doing REW+rePhase each time I change my speaker or sub placement, adjust toe-in, change room furnishings etc. (all of which I do a fair bit), would be an excessive amount of effort. Dirac, the more I’m familiar with it and know how to correctly use it, streamlines this enormously.

Regarding Schroeder or not, for each measurement and cross-over point in Dirac I always generate three filters – one with a curtain at 250Hz, one at 500Hz and one full range. The differences between them are not subtle, but always confined to tonality. There are no weird phase effects or diminished soundstage and imaging with the full-range filter. I tend to select the filter that sounds best with the given program material.

With REW, on the other hand, I’ve been known to go overboard (before I learned to look at group delay etc). I got my measured frequency response ruler-flat, but the music sounded like crap. Dirac seems to shield me from that mistake.

I got more or less the same experience with Mathaudio room EQ. Only big difference between Dirac REW (beeing a tech dummy) is Mathaudio's extreme simplicity. Almost impossible to do anything wrong. An with foobar2000 it's for free remarkable.