Final REW measurements, opinions?

[Davide]

Dear users,

I state that my system is the one in the signature.

I've recently been struggling a bit with Dirac Live due to some issues related to it. I discovered that when measuring at high sample rates the IR it detected was completely wrong, although it sounded satisfactory.
Taking measurements with REW I actually detected some anomalies, so I repeated everything at 44khz.
I also found that the asymmetry in the measurements causes a slight but perceptible difference towards the high frequencies, as Dirac compensates the gain based on these calculations. In fact, at large angles the correction should be limited in band, but since I needed to equalize I preferred to use the Dirac filter directly.

Anyway, I'm quite happy with the result. With the new measurement there is definitely more focus, so Dirac is doing his thing now.

There remain some aspects to improve, such as cancellations between 100 and 200hz, curious GD peaks and also a high GD in the subwoofers, but also high RT60 above 300hz.

Unfortunately I am limited in terms of speaker positioning, this is their definitive position, and in terms of treatment of the room I wouldn't know what to do without obtaining the dissent of my wife and son.
I am attaching a photo of the room (the subs are hidden in the TV cabinet and the door is a grid with a fine wood effect film applied).

I'm attaching some measurements.

Tell me yours!

Thank you

PS. I can't evaluate the IACC calculation... what do you think it is?

 

[Keith_W]

The IACC is a measure of how similar left and right speaker impulses are to each other. What you want is early IACC to be as high as possible (1 = exactly the same). Values above 0.85 are good, above 0.9 are very good. I don't know what values REW uses for "early" and "late" and I could not find it with a quick examination in REW's online manual.

Your early IACC is very poor. Everything from 125Hz up is less than 0.7 meaning that the sound from your left and right speakers are not identical. The obvious reason is how you have placed your speakers. DSP can help somewhat, but ultimately physical problems need physical solutions. If you are unable to find a way to place them better in your room, there is nothing DSP can do to help.

Disclaimer here: I am used to reading IACC in Acourate, which expresses it in % (i.e. > 90% is very good, etc). I am assuming that REW has normalized IACC to 1. If this assumption is wrong, throw everything I said above out the window.

The other obvious problem is that your T20 is very high ... between 600-700ms. You should target between 250-500ms. This means that an impulse will take about 0.6 to 0.7 seconds to decay by 20dB. Again, there is nothing DSP can do about this. You could fix it with more furniture, wall tapestries, thick curtains, room treatment, etc.

The step response looks OK, apart from the very long ringing. Again, not a DSP issue.

In summary:
- Dirac looks like it has done a decent job given what it had to work with.
- Speaker symmetry and placement needs rethinking.
- Excessive reverberation and ringing in the room. Needs work done.

Ultimately it may be better to ignore the measurements and decide whether you like what you hear or not. We humans are very good at adapting to listening rooms, even when the microphone says evil things.

Sorry. I realize this may not be what you want to hear. Good luck.

 

[AnalogSteph]

Looking at the last picture:

Holy mother of floor reflections. Somebody could really do with a nice thick carpet between the table and the TV cabinet. The bare tiled floor is certainly a major contributor to the high reverb times, I know that from my parents. The general lack of carpets and curtains is not helping matters.

Why is the TV cabinet squeezed into the corner that much? I would place it at least one row of tiles over to the right.

Is there a wall behind the sofa, and if so, what does it look like? Walls a short distance behind you tend to be a real trouble spot and generally need heavy treatment.

 

[Davide]

Thanks for the comments!

Why is the TV cabinet squeezed into the corner that much? I would place it at least one row of tiles over to the right.

Is there a wall behind the sofa, and if so, what does it look like? Walls a short distance behind you tend to be a real trouble spot and generally need heavy treatment.

I too would move the TV cabinet to the right but then it is no longer possible to open the dark brown cube on the wall...
That damn low wall that appears on the left seems made just to break the balls... but I am rented and cannot remove it.

Behind the sofa there is a wall with a 130x65cm painting and nothing else.
Not the best this too...

 

[DWPress]

How about putting the left speaker on "damn low wall" and moving the right out on a stand away from the wall? Get some absorption like a nice thick cozy carpet for starters.

As mentioned, Dirac has done its best with what you've presented to it.

 

[Davide]

How about putting the left speaker on "damn low wall" and moving the right out on a stand away from the wall? Get some absorption like a nice thick cozy carpet for starters.

As mentioned, Dirac has done its best with what you've presented to it.

It might be a good idea to improve some aspects, but I cannot place a speaker on a stand with my 3 year old son

And the carpet is definitely something I had thought of, but they are an honest dirt cage.

 

[Davide]

I did a curious experiment.
I redid the Dirac measurements, choosing Wide Imaging Setup, but measuring only the center point, plus two side points with the point-add function (therefore not at those foreseen positions).
This resulted is incredibly better perceived quality, despite having the same frequency and time response. Very similar to the effect of the speakers without Dirac applied. More life like.

 

[napfkuchen]

I redid the Dirac measurements, choosing Wide Imaging Setup, but measuring only the center point, plus two side points with the point-add function.

It does not matter which imaging you choose for the Dirac algorithm. The only difference is the number of available measurement positions.

 

[Davide]

It does not matter which imaging you choose for the Dirac algorithm. The only difference is the number of available measurement positions.

I am not entirely sure of this as for the same measurement points I have generated the filters with different setups (using the function to keep the previous measurements) and the perception is different. But in any case this is not very important ...

The strange thing is that few measuring points should provide a stronger correction for those certain positions, as there is less variance.
In fact, if I produce a filter with only one measuring point, the subsequent measurement in REW shows an extremely flat line at that point (obviously more alterations will be present elsewhere).
And the perception is just as good, but the consistency of the low frequencies is bad in the listening area.
With 3 points it improves the LF consistency in the area and that uncorrected speaker life like effect is maintained.
Increasing the number of meas points would seem to reduce the size of the soundstage, despite this it should mean a less strong correction.

Probably the algorithm is complex and the way it takes into account the number of measurement points is variable. Perhaps the points added to those provided by the setup are taken into account otherwise.

I specify that I am testing the full band correction.

 

[napfkuchen]

There is no 100% working method for the measurements. Dirac's guide on this does not help at all, in fact it is complete nonsense. Judging from other sources, about 6 measurements at a distance of about 20-30 cm seem to make sense. But ultimately, with the numerous measured values, it is your own listening impressions that count.

 

[Keith_W]

Dirac's guide on this does not help at all, in fact it is complete nonsense.

OMG! I didn't realize how bad it was. What a joke!!!

 

[Davide]

There is no 100% working method for the measurements. Dirac's guide on this does not help at all, in fact it is complete nonsense. Judging from other sources, about 6 measurements at a distance of about 20-30 cm seem to make sense. But ultimately, with the numerous measured values, it is your own listening impressions that count.

This is a bit frustrating. In fact, every time I remake the measurements with Dirac, even with the same setup, I get a different sound. Sometimes the soundstage is wider, intelligible, other times it is sterile, collapsed. Sometimes the phantom center is present but backward, other times the sound seems to come out exactly from the cone of the speaker. Sometimes the image is shifted left or right Even if the levels are exactly the same (it will be some precedence effect). Other times the bass have a good slam, other times they are inconsistent.
I also noticed that pointing the microphone up or down slightly changes the high frequencies, especially in the reflections.
This is why I started experimenting with alternative measurement setups like the one I explained earlier.

 

[Keith_W]

The problem with Dirac is that it removes too much control from the user. You only have control over the target curve and nothing else. It works most of the time for most people, but sometimes you need to see what is being done and make your own decisions.

 

[levimax]

This is a bit frustrating. In fact, every time I remake the measurements with Dirac, even with the same setup, I get a different sound. Sometimes the soundstage is wider, intelligible, other times it is sterile, collapsed. Sometimes the phantom center is present but backward, other times the sound seems to come out exactly from the cone of the speaker. Sometimes the image is shifted left or right Even if the levels are exactly the same (it will be some precedence effect). Other times the bass have a good slam, other times they are inconsistent.
I also noticed that pointing the microphone up or down slightly changes the high frequencies, especially in the reflections.
This is why I started experimenting with alternative measurement setups like the one I explained earlier.

I recently purchased DIRAC with the subwoofer support and could not get it to work consistently.... every time I measured I got a different result despite have the Mic in the same places. It also did crazy and inconsistent things with the sub timing. I have a feeling it has something to do with the inherent timing issues with a USB Microphone. I purchased a regular Mic and used it with an interface with a loopback with REW to create room correction filters and set the sub timing based on these measurements and it sounds very nice and everything makes senses and corresponds to physical distances as expected.

See link to this thread on the USB timing issue, there are some high powered people that responded including the creator of REW and VituixCAD explaining these issues that I had no idea about. https://www.audiosciencereview.com/...mic-unsuitable-for-timing-measurements.53175/

 

[Davide]

right. In fact, even in the thousands of tests I have never been able to obtain a correct alignment of the low frequencies.
The overall GD is always higher than the original one.
I also opened a ticket to Dirac and they investigated the case, but months have passed. Their initial reply was that Dirac's goal is to reduce all GDs where possible...
But in fact it makes things worse ...

 

[Davide]

Look at this attempt with 12 measuring points.
The graph is titled Group Delay but actually only the Excess Phase is displayed (Minimum phase was estimated by REW with the full bandwidth sweep).
So, the excess phase is huge (up to 100 ms as you can see), and the most incredible thing is that without Dirac it is almost 0 ms (using a main speaker as an acoustic timing ref).
I don't understand if this is due to the search for the best frequency response, or if it is deliberately introduced to create greater bass presence (I have seen that other people get the same time worsening).
This phase shift seems quite audible to me, but I have not conducted an appropriate test to say it.

 

[levimax]

Interesting graph. In my case the DIRAC sub timing was about 125 ms off (The correct amount of adjustment relative to the mains and the LP was to delay the mains 4 ms) so that is huge and I found it very audible. DIRAC did a good job on the FR of the mains if I measured them without the sub but was counterproductive for sub integration. Diving into manual acoustic measurements and timing issues using a USB Mic and I can understand why. Another issue I discovered is that my subs (SVS 30000) use DSP which adds 6 ms delay and some pre-ringing. To me it seems it would be impossible for DIRAC to take into account all the possible scenarios of different rooms, subs, sub placement and USB measurements variations and that they should have stuck with just FR corrections which they seem to have down.

 

[Davide]

Interesting graph. In my case the DIRAC sub timing was about 125 ms off (The correct amount of adjustment relative to the mains and the LP was to delay the mains 4 ms) so that is huge and I found it very audible. DIRAC did a good job on the FR of the mains if I measured them without the sub but was counterproductive for sub integration. Diving into manual acoustic measurements and timing issues using a USB Mic and I can understand why. Another issue I discovered is that my subs (SVS 30000) use DSP which adds 6 ms delay and some pre-ringing. To me it seems it would be impossible for DIRAC to take into account all the possible scenarios of different rooms, subs, sub placement and USB measurements variations and that they should have stuck with just FR corrections which they seem to have down.

It will also be a generic and non-customizable tool, but if the speakers/subs have low GD uncorrected, why should they reach 125 ms when corrected?
I know that Dirac to keep latency low does not use FIR at low frequencies.
Probably when a lot of gain with IIR is needed, as to improve the cutoff at the low end, the effect is a large phase deviation.
But I'm not sure this is the cause.
Perhaps it has to do with the variance between the various measuring points and the goal of having a good frequency response.

 

[levimax]

Looking at the measurement variations inherent with a USB Mic https://www.avnirvana.com/threads/acoustic-phase-measurements.13300/post-99941 and then taking multiple measurements all with random timing errors it is no surprise DIRAC struggles.

 

[Basic Channel]

The IACC is a measure of how similar left and right speaker impulses are to each other. What you want is early IACC to be as high as possible (1 = exactly the same). Values above 0.85 are good, above 0.9 are very good. I don't know what values REW uses for "early" and "late" and I could not find it with a quick examination in REW's online manual.

There is an explanation here

The IACC (Inter-Aural Cross Correlation) panel is used to calculate the early, late and full IACC values for a pair of binaural measurements. The zero time for the measurements is taken as the earlier of the two measurement IR start times. The division between early and late is 80 ms later. The end of the "late" period is 500 ms or the end of the Schroeder integral (where the integral falls into the noise floor), whichever is later. REW does not change the alignment of the measurements, if they need to be aligned the Cross corr. align option may be suitable. The measurements are filtered with one octave, zero phase bandpass filters of the chosen order.