Dirac with and without REW results

[kyuu]

In order to have different target curves for the left and right speaker they would have to be in different groups. You can do that by just dragging one or the other out of Group 1 and into the blank space. However, I can't think of any reason you would want or need to do that. All your speakers, including the subwoofer, should be using the same target curve. You only break the subwoofer out because you typically want to curtain it differently than the mains.

Something is definitely weird, no way should you want Dirac to correct to a curve with peaks and dips like that. What crossover settings (frequency, filter if the Bluesound exposes that) are you using? Is the SVS's DSP doing anything?

 

[Jaxjax]

hey so i have been playing around..shoot me now and tell me WTH have you done..but ok so attached is my latest REW showing L/R before and after Dirac and also the L+R after dirac now. I have also attached the dirac sub curve SS and the Dirac live L speaker curve which i have manually moved the points around to achieve what is now showing in REW. This manual curve looks so wrong with the way i have moved the curved into a more zig zag postion but it seemed to smooth the response when i measured in REW afterwards. Please tell me if this is ludicrous and i should immediately put it back to a smooth curve in dirac Also, is there a way to manually move measurement points individully for both the left and right speakers? If i move the points in the L speaker then it replicates in the R

I unzipped your REW. Have you tried to curtain that off to 300 & let Dirac do its thing on down & adjusting your bottom end shelve to taste .? Is your system harsh without Dirac or something.? Also do you have 2034 data on those speakers.?

 

[regan]

I don't use Dirac.

What do you use and why do you prefer it to Dirac?

 

[Keith_W]

hey so i have been playing around..shoot me now and tell me WTH have you done..but ok so attached is my latest REW showing L/R before and after Dirac and also the L+R after dirac now. I have also attached the dirac sub curve SS and the Dirac live L speaker curve which i have manually moved the points around to achieve what is now showing in REW. This manual curve looks so wrong with the way i have moved the curved into a more zig zag postion but it seemed to smooth the response when i measured in REW afterwards. Please tell me if this is ludicrous and i should immediately put it back to a smooth curve in dirac Also, is there a way to manually move measurement points individully for both the left and right speakers? If i move the points in the L speaker then it replicates in the R

Whatever you have done, it is an improvement.

Left (blue) + right (yellow) post Dirac, with a target curve superimposed. This isn't as deficient in bass as it was before, but it is still deficient. It would be nice to get maybe 2-3dB more bass below 60Hz.

Your right speaker is noticeably more lumpy than the left, in particular there is a huge peak at 300Hz which is about +7dB. I don't know if it is because you are using that weird looking zig-zag target curve, but you are right - it is ludicrous. You should immediately put it back to a smooth curve, and use the same curve for left and right.

Anyway, let's look at more measurements and comment on them.

On the left we have the IACC (inter-aural cross correlation) before Dirac compared to after Dirac. The IACC compares how similar the left and right channels are. The more similar the two channels, the better your stereo image will be. 1 is perfect correlation, 0 is no correlation. You want the scores to be as close to 1 as possible. Most typical rooms are about 0.7, "good" is 0.8 and above, and "excellent" is 0.9 and above. "Early" is the direct sound of your speaker (which you can manipulate with DSP), and "Late" is the reverb or decay of your room (which you manipulate with room treatment).

Let us focus on "early". We can see that Dirac has improved your IACC somewhat - at 63Hz it has gone from 0.606 to 0.873. There are slight improvements elsewhere, e.g. at 1kHz it has gone from 0.731 to 0.799. But that IACC at 4kHz and 8kHz is dismal. We can see that it was already terrible before Dirac, and there isn't much Dirac can do about it. Either your speakers were poorly manufactured and you got a really bad copy where the tweeters don't match, or you have positioned them poorly. I think the latter is more likely.

This is the first 200ms of your step response. I don't normally zoom out so much, but I did it to show you some serious problems. On the left, we have "before (red) and after (green)" for the left speaker. On the right we have post-Dirac for left and right speakers.

FYI, this is my step response zoomed out to the same 200ms time scale so that you have an idea what it should look like:

After the main impulse at 0ms, all waves that follow should be lower in amplitude than the main. If it is higher than the main, then either the time alignment was done poorly, or reflections are arriving that are louder than the main impulse. In this case, I think that Dirac has done a very poor job of subwoofer time alignment. If that huge peak on the graph on the right is the subwoofer's main impulse, then your sub time alignment is off by about 30ms. This is huge.

The reason your main impulse does not look as clean as mine is because yours is heavily contaminated by reflections.

Since we are talking reflections, this is the Energy-Time Curve (ETC) of your left (blue) and right (yellow) speakers post Dirac. The rule is - in the first 20ms, all reflection peaks should be a minimum of -15dB to the main impulse (the main impulse is that big peak at 0ms). Also, each peak should be symmetrical between left and right (implying the speakers have been set up symmetrically in the room). I have drawn in the -15dB target and put green arrows on all the peaks.

I realise I am giving you a lot of information. You likely won't be able to understand and fix everything at once. But I suggest you start by taking a good look at your system and how you have positioned your speakers. For a start, there are a lot of very early and very loud reflections, and they are not symmetrical between left and right. Before you even begin to DSP, you need to position your speakers properly.

 

[Keith_W]

What do you use and why do you prefer it to Dirac?

I use Acourate. IMO Dirac has too much automation and removes too much control from the user. Dirac is easier to use since it automates a lot of things which require manual interpretation, but I don't trust an algorithm to interpret things for me. I prefer to look at it myself and decide. Acourate only does what it is told to do, it does not "think" and it certainly does not think it knows better than the user. So the user can also fix it.

 

[Firewalk61]

In order to have different target curves for the left and right speaker they would have to be in different groups. You can do that by just dragging one or the other out of Group 1 and into the blank space. However, I can't think of any reason you would want or need to do that. All your speakers, including the subwoofer, should be using the same target curve. You only break the subwoofer out because you typically want to curtain it differently than the mains.

Something is definitely weird, no way should you want Dirac to correct to a curve with peaks and dips like that. What crossover settings (frequency, filter if the Bluesound exposes that) are you using? Is the SVS's DSP doing anything?

I have the SVS set to cross over at 70 phase at 0 and level at -10 (interestingly the volume level on the sub goes from 0db to -60db. Not sure if the is is normal.
The Icon just has options for the crossover which is at 70 and some replay gain. There isn’t a volume button there so assumed it did it automatically and I just set the sub level on the sub itself

 

[Firewalk61]

I unzipped your REW. Have you tried to curtain that off to 300 & let Dirac do its thing on down & adjusting your bottom end shelve to taste .? Is your system harsh without Dirac or something.? Also do you have 2034 data on those speakers.?

I haven’t tried yet. I’ll give it a go..here is the data sheet for the speakers

 

[Firewalk61]

Whatever you have done, it is an improvement.

View attachment 440006

Left (blue) + right (yellow) post Dirac, with a target curve superimposed. This isn't as deficient in bass as it was before, but it is still deficient. It would be nice to get maybe 2-3dB more bass below 60Hz.

Your right speaker is noticeably more lumpy than the left, in particular there is a huge peak at 300Hz which is about +7dB. I don't know if it is because you are using that weird looking zig-zag target curve, but you are right - it is ludicrous. You should immediately put it back to a smooth curve, and use the same curve for left and right.

Anyway, let's look at more measurements and comment on them.

View attachment 440010View attachment 440009

On the left we have the IACC (inter-aural cross correlation) before Dirac compared to after Dirac. The IACC compares how similar the left and right channels are. The more similar the two channels, the better your stereo image will be. 1 is perfect correlation, 0 is no correlation. You want the scores to be as close to 1 as possible. Most typical rooms are about 0.7, "good" is 0.8 and above, and "excellent" is 0.9 and above. "Early" is the direct sound of your speaker (which you can manipulate with DSP), and "Late" is the reverb or decay of your room (which you manipulate with room treatment).

Let us focus on "early". We can see that Dirac has improved your IACC somewhat - at 63Hz it has gone from 0.606 to 0.873. There are slight improvements elsewhere, e.g. at 1kHz it has gone from 0.731 to 0.799. But that IACC at 4kHz and 8kHz is dismal. We can see that it was already terrible before Dirac, and there isn't much Dirac can do about it. Either your speakers were poorly manufactured and you got a really bad copy where the tweeters don't match, or you have positioned them poorly. I think the latter is more likely.

View attachment 440011View attachment 440013

This is the first 200ms of your step response. I don't normally zoom out so much, but I did it to show you some serious problems. On the left, we have "before (red) and after (green)" for the left speaker. On the right we have post-Dirac for left and right speakers.

FYI, this is my step response zoomed out to the same 200ms time scale so that you have an idea what it should look like:

View attachment 440012

After the main impulse at 0ms, all waves that follow should be lower in amplitude than the main. If it is higher than the main, then either the time alignment was done poorly, or reflections are arriving that are louder than the main impulse. In this case, I think that Dirac has done a very poor job of subwoofer time alignment. If that huge peak on the graph on the right is the subwoofer's main impulse, then your sub time alignment is off by about 30ms. This is huge.

The reason your main impulse does not look as clean as mine is because yours is heavily contaminated by reflections.

View attachment 440019

Since we are talking reflections, this is the Energy-Time Curve (ETC) of your left (blue) and right (yellow) speakers post Dirac. The rule is - in the first 20ms, all reflection peaks should be a minimum of -15dB to the main impulse (the main impulse is that big peak at 0ms). Also, each peak should be symmetrical between left and right (implying the speakers have been set up symmetrically in the room). I have drawn in the -15dB target and put green arrows on all the peaks.

I realise I am giving you a lot of information. You likely won't be able to understand and fix everything at once. But I suggest you start by taking a good look at your system and how you have positioned your speakers. For a start, there are a lot of very early and very loud reflections, and they are not symmetrical between left and right. Before you even begin to DSP, you need to position your speakers properly.

Thanks Keith. I’ll try to process the information you have provided and see what I need to do. Do you think the sub being out of phase would do this? My room isn’t that bad a room( I don’t think) not sure what’s going on. I had initially set the speakers up using the rule of 1/5 then played about with them a little more. I have them set from the side walls 70cm 1/5. And initially had them off the back walls at 80cm, but last night found that bringing them right in closer to the back walls 60cm got better measurements..but then this could have caused the above issues? Iv attached a few pics of the room for reference, listening position is at 4m from speakers.

 

[Keith_W]

Please show us a photo of your microphone and where you place it when you take measurements. The sofa on the left is likely creating some reflections. If it needs to be there, then don't worry about it.

Yes, phase issues from subwoofers can create these problems. But Dirac should be able to fix that. I recommend you turn OFF DSP on your sub (if any), then set phase to 0deg. Lower the volume of your sub by about 8dB. You don't want DSP to cut too much volume from the sub.

 

[Firewalk61]

Please show us a photo of your microphone and where you place it when you take measurements. The sofa on the left is likely creating some reflections. If it needs to be there, then don't worry about it.

Yes, phase issues from subwoofers can create these problems. But Dirac should be able to fix that. I recommend you turn OFF DSP on your sub (if any), then set phase to 0deg. Lower the volume of your sub by about 8dB. You don't want DSP to cut too much volume from the sub.

Ok I’ll position soon and send pic..1 thing to note which I have always found strange..is when I connect the sub direct to the amp pre out -10 the base is just crazy crazy loud to point it’s just rumbling..when I connect the sub to the icon sub out..at -10 it is prob 90% quieter than from when plugged into the amp..this is with the sub cross over at 70 and the icon same, no other DSP on with the icon..always found that strange

 

[Old_School_Brad]

I believe it's difficult to directly compare Dirac measurements with those from REW. The reason is that we don’t fully understand how Dirac weighs measurements taken at different positions to reach its target. For example, Dirac typically uses 11 measurements, with the std. settings, if I recall correctly, at various points around the listening area. So, how would you replicate those measurements manually? How should you weight the four measurements to the sides of the main listening position compared to the center? And what about the outermost measurements -how should they be weighted?

We really can't know for sure, can we?

While conducting a MMM across the entire area might give some insight, what conclusions can we draw from it?

I’m not entirely convinced that comparing in this way is all that valuable.

 

[Firewalk61]

Please show us a photo of your microphone and where you place it when you take measurements. The sofa on the left is likely creating some reflections. If it needs to be there, then don't worry about it.

Yes, phase issues from subwoofers can create these problems. But Dirac should be able to fix that. I recommend you turn OFF DSP on your sub (if any), then set phase to 0deg. Lower the volume of your sub by about 8dB. You don't want DSP to cut too much volume from the sub.

Mic position central to the speakers

 

[Keith_W]

This is why you are getting so many reflections. You may have a boom tripod, but you are treating it like a mini-tripod. Listening so close to the rear wall is not ideal. I realise you may not have much choice in the matter so I guess it is what it is.

It is this kind of situation where I wonder if Dirac is smart enough to avoid doing something inappropriate. Like attempt to correct all the reflections.

I can tell from the style of architecture that you are in the UK?

 

[Firewalk61]

View attachment 440071

This is why you are getting so many reflections. You may have a boom tripod, but you are treating it like a mini-tripod. Listening so close to the rear wall is not ideal. I realise you may not have much choice in the matter so I guess it is what it is.

I can tell from the style of architecture that you are in the UK?

Yes that the listening position which I cannot change, the tests were done whilst curtains closed..not that that would make any difference I don’t think. So are we saying I can’t do anything about the problems?

 

[Keith_W]

Yes that the listening position which I cannot change, the tests were done whilst curtains closed..not that that would make any difference I don’t think. So are we saying I can’t do anything about the problems?

Reflections are an acoustic problem, not a DSP problem. If you have a copy of Toole, you can read what he says in Chapter 7.6 about the audible effect of early reflections (in a nutshell, it smears the sound and reduces clarity). Reflections need to be delayed and attenuated. The only way to delay reflections is to sit further away from reflective surfaces. You can attenuate reflections by sitting further away, or by using acoustic absorbers. Or you can choose to live with it. Those are the only options. Living with it is not so bad, lots of people have no choice.

Reflections only become a DSP problem if DSP inappropriately tries to correct it. This is why I don't like Dirac, I don't think it gives you enough control over what to correct and what to leave alone. With Dirac, I believe there is an option to correct up to a certain point. I suggest you cut off correction above 200Hz or so.

As for the other issues - poor subwoofer time alignment, bad step response, etc. - you might be able to do better if you used a DSP with more manual control. But those are very difficult to use and require a lot of study and experience. For the time being, I suggest you work with what you have and learn what those measurements mean. In the future, if you get the itch, you can try more manual DSP software.

 

[kyuu]

I have the SVS set to cross over at 70 phase at 0 and level at -10 (interestingly the volume level on the sub goes from 0db to -60db. Not sure if the is is normal.
The Icon just has options for the crossover which is at 70 and some replay gain. There isn’t a volume button there so assumed it did it automatically and I just set the sub level on the sub itself

Correct me if I'm wrong, but it looks like you have a crossover enabled in both the the Bluesound and in the SVS sub? You don't want that. Use one or the other, not both. I would use the Bluesound's. The SVS also defaults to a 12dB/octave filter for some reason it looks like. You should be using at least 24dB/octave if available.

 

[Firewalk61]

Correct me if I'm wrong, but it looks like you have a crossover enabled in both the the Bluesound and in the SVS sub? You don't want that. Use one or the other, not both. I would use the Bluesound's. The SVS also defaults to a 12dB/octave filter for some reason it looks like. You should be using at least 24dB/octave if available.

Hi yes sorry i turned the crossover off on the sub and left icon crossover on, I will try changing the octave to 24, as it is as you say 12

 

[Old_School_Brad]

What are you trying to achieve by doing these measurements?