A room correction case

[serial]

Hi,

I'm a newbie in the acoustics world, just started discovering everything and experimenting. For now, I have Denon 3800 with Audyssey XT32 and 2 tower speakers. Central speaker and a subwoofer are on the way, so I will redo the experiments below when they arrive. Also ordered UMIK-1 for REW and Dirac.

I have an irregularly shaped room (see the picture) and the following measurement taken with the Audyssey app, visualized in Ratbuddyssey (don't have the mic for REW yet). I took the measurement several times, so it should be correct. It shows a huge spike around 70 Hz and sever decline after 2kHz. I tried to move speakers out of the corners, tried several positions with no real difference until I move them completely out of the niche. The 2nd picture shows the measurement with the speaker placed at the room center, and it looks fine, I think.

I hope the room can be corrected with either Audyssey or Dirac, but what else can / should be done (some kind of room treatment, I guess) to help them? Because the 1st measurement looks pretty bad. Placing speakers not in the niche is the last option I would try, only if the room cannot be corrected.

Thanks.

The room:

Speakers in the niche, before correction:

Speakers out of the niche, before correction:

 

[Daverz]

The decline after 1 kHz is to be expected and looks very good as far as linearity, actually.

 

[DWPress]

The middle of the room looks so much better but that's not where you want your speakers is it? If you are going to use Audyssey or Dirac to make corrections it will most likely give you a smoother result despite the niche location. Not familiar with Audyssey but Dirac and most other automated systems will trim the peaks and make frequency boosts full range or up to a specified frequency (usually your rooms Schroeder frequency +/- 500Hz) to give you the sort of room curve you desire.

The graphs you show us above don't look terrible but also not very detailed. You'll have to wait until the rest of your speakers and Umik arrive to find out the true story about what's going on with the speakers interaction with the room. Sub location will be tricky but, if also placed in the niche, could gain even more bass SPL matching that 70Hz hump which DRC can tame and shelf up your higher frequencies.

No way to know until you get your kit and measure the space and drivers properly. I would at least try moving the speakers towards the MLP to the edge of the niche and see what kind of differences show up.

 

[davidc]

Your response with the speakers out of the niche is excellent. With them in the niche, it's to be expected. The decline after 2k is not "severe". It's only about 5dB. Audyssey and Dirac Live are very similar. No huge differences, and should be able to tame that peak. You didn't say exactly were the speakers are...if they are against the wall, try moving them at least a foot away.

 

[ozzy9832001]

You'd really need to test both speakers individually. My guess is you'll see some differences between them due to side wall distances at MLP being different. This could (key word) affect stereo imaging. You could easily correct the bass hump with EQ or RC software. Most people prefer a downward slope in their FR...it sounds more natural, so the dip may not even be audible or it may be more pleasing. We are more sensitive to those frequencies.

 

[serial]

Got UMIK-1 and Dirac and did quite a few measurements. For now, the setup is 4.0 with some very old surround speakers (will be replaced with proper ones soon), so I don't include them in the graphs.

First of all, SPL for both L and R before correction look almost the same. I tried many different positions and, from REW, concluded that toe in is better that straight position, and also of course the further from walls the better.

Correction to flat with both Audyssey and Dirac shows identical curves with Audyssey +1db higher at frequencies above 6k. From that, being a complete novice in the field, I can conclude that AVR has a fixed number of EQ bands and both Audyssey and Dirac set them to identical positions.

The surprise for me was that not only does Audyssey Dynamic EQ increase bass and treble, but it also: 1. adds +5db to surround speakers (Dyn EQ curve for surrounds is the same as for fronts but just 5db higher), 2. adds some kind of surround sound effect. For me, it is a killer feature because it just sounds much nicer and I'm not an audiophile purist. I measured the exact curve from Audyssey Dynamic EQ (see the graph), applied it to Dirac and still the surround effect is not there, so Audyssey wins.

I will repeat measurements after the rest of my speakers arrive (will be 5.1) but for now, I'm not impressed with Dirac despite all the hype. Maybe it will integrate my upcoming subwoofer better than Audyssey, let's see.

For now, the question - based on the graphs, what can I potentially do to improve the sound in the room?

Part of the room:

Left and right before correction:

Left corrected to flat with both Audyssey and Dirac (blue line):

Filtered IR left speaker uncorrected, after correction the line drops down by 30%:

Waterfall left speaker uncorrected:

Waterfall left speaker with Audyssey flat:

FYI, from what I measured, this is the Audyssey Dynamic EQ curve - front (the line below) and surround speakers (the line above):

 

[OCA]

what can I potentially do to improve the sound in the room?

Both automated correction systems have flaws. For best results, I'd suggest you cut off correction at around 200Hz since they both blindly apply frequency magnitude correction on octave band smoothed response at high frequencies to match a target curve which will degrade the sound for all but the worst speakers as beam directions and peak energy times are completely ignored. With Audyssey, you might benefit from taking only 3 measurements (the minimum required) all at MLP without moving the mic since it has a very ancient way of spatially averaging measurements at different mic positions and gets the average speaker response quite wrong.

You have correctly detected the DEQ effect on surround speakers and you're the only person I have ever seen to enjoy that effect and you helped me understand why Audyssey is doing it

 

[ozzy9832001]

I'd second the advise you've already gotten. For your situation, I'd cut off around 200hz where that last hump is. I'd also apply a +3-5dB shelf from 100 - 120hz and below (to taste). The balance will be much better, especially if you listen at lower volumes.

 

[serial]

Thanks everyone for suggestions, really appreciate it. I cannot limit correction to 200-300 Hz because I need increased highs, see details below.

Now I have 5.1 setup with Kef Q950, Q650c, Q150, SVS SB-1000 Pro, receiver Denon AVC-3800H, waiting for 4 Atmos speakers to arrive next month. After several weeks of experimenting with speakers placement and room correction, I finally managed to come to the sound I wanted. The thing is that, from the very beginning, I was suffering from lack of clarity. Everything sounded kind of muffled and I was trying to understand is it something with the speakers or my ears - tried different speakers but still not sure that it is though. What I did to radically increase clarity to the level I need:

1. Due to lack of space and not hearing much improvement from moving front speakers around, I finally left them where they were, placing just in a visually appealing position.
2. After doing a lot of comparison, I must admit that Dirac sounds so much better than Audyssey so that it seems to be one of the best investments, BUT I had to do something strange which I've never read on forums:
3. I did Dirac measurement with the microphone turned 180 degrees (the opposite direction) from speakers, with 0 deg calibration file. This alone increases highs by a lot.
4. Target curve in Dirac: Harman 6db (found the curve somewhere on forums) + 0 level from 400 to 7kHz + sharp increase from 8.5kHz. Tried multiple curves, increasing from 8.5kHz was the best for my ears.
5. Increased surrounds and subwoofer levels by +5dB.
6. Tone / treble increase to taste. Dirac live allows it while Audyssey doesn't.
7. Toe in almost straight to MLP.
8. Replaced speaker cables. Of course, I've read about snake oil cables but when I replaced my 10 euros / 10 meters 1.5 mm Hama cable CCA with 40 euros / 10 meters Oehlbach 2.5 mm OFC, the difference in treble was audible. Maybe the snake oil effect only applies when the price goes beyond 100 euros or so. Obviously, the moral is not to buy 10 euros cable for connecting 1-2k worth speakers, stupid thing.

For comparison, I have slot 1 in Dirac with the measurement done as traditionally suggested - mic to the ceiling, 90 deg calibration file, Harman target curve for bass and flat in highs, and slot 2 with what I described above, and slot 2 sounds much better and clearer.

Also a question - people write that you should correct to Schroeder frequency because room only has effect below it, but why then I have around 5-7db decrease from 1kHz to 10 kHz when I measure with REW at MLP and I don't have the decrease with the mic right next to the speaker - isn't it room effect? I have a very thick (15 mm) carpet on the floor, might be the reason.

Two more side points. I believe that the dynamic volume feature available in Audyssey is so valuable that it should be available on top of any room correction as well as without it. I also ended up to be disappointed in DEQ after extensive comparison with it / without it / with Dirac.

 

[ozzy9832001]

Also a question - people write that you should correct to Schroeder frequency because room only has effect below it, but why then I have around 5-7db decrease from 1kHz to 10 kHz when I measure with REW at MLP and I don't have the decrease with the mic right next to the speaker - isn't it room effect? I have a very thick (15 mm) carpet on the floor, might be the reason.

Two more side points. I believe that the dynamic volume feature available in Audyssey is so valuable that it should be available on top of any room correction as well as without it. I also ended up to be disappointed in DEQ after extensive comparison with it / without it / with Dirac.

Well there's a lot of reasons, but I'll explain 2 that will apply.

From say 1Khz - 5 or 6khz we perceive these sounds to be louder. Humans are most sensitive in this range. Most alarms, sirens, things of that nature are in this range. So, even though they maybe lower, you may actually perceive them just as loud. This would be a situation where to taste is best.

The Xover to the tweeter is probably around 2.5khz to 3khz. Sound will beam from the tweeter. It's completely direction. So, if the tweeter isn't pointed exactly at the MLP, you will get that drop.

Experiment with toe in to "correct" the issue. Again, it is one to taste. Changing the toe in will change the points of reflection for everything else. This maybe desirable, it might not be.

This is why correcting purely direction sound is bad. You will usually end up with a very bright system because the mic is positioned in 1 spot, but we hear with 2 ears. You're best best to get a good quality, overall representation would be the MMM, but even that has some limitations, as your going to be scanning areas that your head or ears will never be in.

 

[keks8430]

This is why correcting purely direction sound is bad. You will usually end up with a very bright system because the mic is positioned in 1 spot, but we hear with 2 ears.

Hi ozzy,

I am not sure about this reasoning.

With our two ears and the processor between we can distinguish somehow between speaker sound and room acoustics.
The single-channel omni-directional mic measurement cannot easily distinguish, it simply sums direct and reflected sounds.
Since the highs have higher directivity, the overall sum of high sounds will be less than those for bass and mid frequencies.
That makes your room curve slope down.
Only then, if you set your target curve to flat, you will end up with a bright sounding system.
That's why you should start with a target curve smoothing the frequencies above Schrodinger (~300Hz).
Or, not correcting above at all, taking care of a continuous gain transition between the two regions.

Hopefully this stands the scrutiny of our forum experts!