Dirac Live Free Trial

[Hlam]

My 14 day trial is almost up and my results have been very poor. Dirac live sounds worst than no eq and much worst than Audyssey XT32.
I think that the problem may be the mike I am using. It is the mike that came with my Omnimic-V2.
This is one of the mikes that is listed as compatible with the free version. Dirac installs the calibration file and the mike test comes back as installed and working properly. The measurement procedure runs flawlessly and once the results are optimised and loaded into the DAP they are streamed to the receiver.
Is there anyone here that has used the Omnimic mike with the free trial version of Dirac? If so what were your results?
I can not rule out human error. I never used a computer as an output device before.
Thanks

 

[Sonnie]

Hmmm... never used that mic or the free trial, but I'd say something is definitely not working correctly. Maybe someone who has can chime in.

Have you contacted DL support?

 

[AudiocRaver]

Dirac Live "sounds worse" in what way? If it is your first experience with the tool, there are many variables to be considered.

 

[AudiocRaver]

Also, I believe that most who have worked with Dirac in a number of rooms would claim that if Dirac "makes a room sound worse," it is probably not being applied in the right way.

1. How does it "sound worse?" Boomy bass? Weak bass? Distortion? Soundstage is scrambled or messed up? Imaging is poor/worse? Frequency response has an "odd" tonality?
2. Are there other possible variables? Speakers got moved a little? Distance or level values got changed somewhere?
3. Have you gone through the Windows video card setup for the video card that has the HDMI output (is that even y our configuration?)
4. What target curve are you using?

 

[Hlam]

Thanks for the input.
The bass sounds almost as good as no eq. It is the vocal and upper midrange that sound off. The imaging and soundstage are slightly off when compared to no eg or audyssey.
The speakers have not been moved.
I reset my receiver to factory defaults, to zero out the distances, and ran Dirac again. I think that this is 15 times I have did the measurements now.
I got the same results. Playback through the Dirac filters sound worst than using no eq.
I am running Dirac directly to my receiver through HDMI using the DAP that comes with the trial version. The measurement tones are sent to all six channels correctly.
I have only used the target curve that Dirac suggested.
It is to bad that the trial version for a PC is only 14 days and not the 30 days that the Mac users get. I will be out of time before I get this figured out.
I know that it is either my equipment or something that I am doing wrong. Dirac should not make my system sound worst.
Thanks again for the help people.

 

[Flak]

@Hlam

we will be pleased to renew your trial if you'll send a request here:
http://helpdesk.dirac.se/index.php?pg=request
I anyhow suggest that you wait for the upcoming completely rewritten version:
https://www.avnirvana.com/threads/dirac-research-is-primed-to-launch-major-update-at-ces-2018.1874/

Flavio

 

[Hlam]

Thanks Flavio.
The problem could be me. I have never used a computer to stream music before and will most likely never again after this trial.
I was intending to purchase a Minidsp DDRC-88A/BA unit. Will these units be upgradable to the new version of Dirac?
I understand if you cannot answer that question. I should send that one to Minidsp in the future.
I will try to add a picture of my measurements comparing no eq. to Audyssey XT32 to Dirac.

 

[Flak]

Hi Hlam,
you're quite right in being unsatisfied with the listening results, the frequency response is much worse.
I suggest that you open a ticket at the link that I gave you above and send your project, we will have a look at it and extend your trial.
With reference to the upgrade of the miniDSP DDRC-88 you'll find the answers by miniDSP itself here:
https://www.minidsp.com/forum/dirac-series-support/13793-dirac-live-2018-2-0-minidsp-support

Flavio

 

[AudiocRaver]

I agree, according to your measurements, it should sound worse.

What it looks like is a measurement taken with both L and R mains active and the mic not quite centered between the two, so there is some cancellation at the upper mids range. But that would give an ugly curve like you see, while sounding much better. In your case, it apparently sounds as bad as it looks.

Check your Dirac configuration and your Windows settings for the HDMI output. Is it possible that your L and R channels are somehow being mixed to mono, so the L goes to L and R and the R goes to L and R?

 

[Hlam]

Hi Hlam,
you're quite right in being unsatisfied with the listening results, the frequency response is much worse.
I suggest that you open a ticket at the link that I gave you above and send your project, we will have a look at it and extend your trial.
With reference to the upgrade of the miniDSP DDRC-88 you'll find the answers by miniDSP itself here:
https://www.minidsp.com/forum/dirac-series-support/13793-dirac-live-2018-2-0-minidsp-support

Flavio

Thanks again.
I will do as you suggested.

 

[Hlam]

I agree, according to your measurements, it should sound worse.

What it looks like is a measurement taken with both L and R mains active and the mic not quite centered between the two, so there is some cancellation at the upper mids range. But that would give an ugly curve like you see, while sounding much better. In your case, it apparently sounds as bad as it looks.

Check your Dirac configuration and your Windows settings for the HDMI output. Is it possible that your L and R channels are somehow being mixed to mono, so the L goes to L and R and the R goes to L and R?

Thanks for the input.
The three measurements were taken at the MLP with the front left, front right, and subs active. The mike was moved while I took the three measurements. If this is a cancellation due to mike position don't you thing that it should show up in all the measurements.
The measurement tones seem to cycle through the system properly.

 

[Hlam]

It may be something in my receiver settings.
I did the measurements with Audyssey turned off. I used the distances and levels that audyssey set.
Just as a test I reset my receiver to factory defaults for the last set of measurements. Distances and levels all at zero. I think that this was wrong. I do not think that this version of Dirac will set these on it's own. Please correct me if I am wrong.

 

[Matthew J Poes]

It may be something in my receiver settings.
I did the measurements with Audyssey turned off. I used the distances and levels that audyssey set.
Just as a test I reset my receiver to factory defaults for the last set of measurements. Distances and levels all at zero. I think that this was wrong. I do not think that this version of Dirac will set these on it's own. Please correct me if I am wrong.

When you measure both the Left and Right speakers at the same time it causes a cancelation like we see here. The fact that it isn't showing up when there is no EQ or Audyssey could be a sign that there is a phase issue that Dirac has attempted to correct. Once corrected then both speakers would be more likely to cause this cancelation.

You show significant comb filtering in the high frequencies in all measurements.

I suggest taking a single measurement of the Left only+sub and right only+sub. I would also carefully check polarity of all speakers.

What kind of speakers do you have? Is there any chance that one of them has a tweeter wired wrong?

I would definitely take up Flaks suggestion and also work with Dirac directly. Your results are not typical. Once you did figure out the problem please report back so we can help others if it ever shows up again.

 

[JStewart]

It may be something in my receiver settings.
I did the measurements with Audyssey turned off. I used the distances and levels that audyssey set.
Just as a test I reset my receiver to factory defaults for the last set of measurements. Distances and levels all at zero. I think that this was wrong. I do not think that this version of Dirac will set these on it's own. Please correct me if I am wrong.

Dirac does set distances and levels. You can see the values its using by clicking on the left pointing arrow next to each filter (right hand side of each filter). Dirac does recommend setting the distances and levels as best as possible before a calibration.

 

[AudiocRaver]

Yep, figured as much. Live and learn, we have all made that mistake at some point, no doubt. Measuring all subs together too see how they work together below 80 Hz is commonly recommended, but measuring mids and highs with a swept sine wave (like REW) through more than one speaker at once will almost always give you nonsense.

 

[Hlam]

I did another Dirac calibration and this time I ran Audyssey using the same mike positions. Except for position # 9 of course.
I got the same results as you can see in the graph.

 

[Hlam]

The speakers are Ascend Acoustics Sierra 2's.
The polarity looks good at the receiver and the speakers. The tweeters seem to be Ok.
I did the measurements that you suggested.

 

[Matthew J Poes]

Have you tried contacting dirac?

Can you try using REW to take a measurement. That will help us see if this is an interaction between the mic and software. The mic that comes with Omnimic, as I understand it, is a UMM-6 that has been tied to the software. It is otherwise identical and as such should work the same as any UMM-6. I know the UMM-6 works fine with Dirac. By using REW you can also send us the MDAT file and we can look for odd behavior.

What would be best, for trouble shooting, would be to do a Dirac correction again, but use just a single measurement point. Put the mic in the center of your listening position and take a measurement. Then wait for it to process that file and move forward with the correction. When you make a correction, it is important that the house curve mostly lies on the response curve. If it lies above most of the response curve, I find that it will try to boost everything and you get poorer results, with most holes where it simply can't boost enough).

If you can do this correction and then measure again with REW doing the Left+sub and Right+sub as separate measurements, that will help. Then upload the MDAT file to your post and we can look at everything to see if there are clues.

Another thought, make sure that the measurement software is using the correction properly. To do so the output device needs to be the DIRAC processor and not your receiver anymore (and in the Dirac software, it needs to be using your receiver output). You also need to be sure that there is nothing on in the soundcard, no processing of any kind. When you measure, try measuring with Dirac turned on and off using the Dirac switch rather than changing the output. This not only preserves the same output level, but helps troubleshoot the problem. During the measurement, keep an eye on the Dirac clip indicator, just in case there is some kind of signal clipping going on. I have made a few correction files that due to their shape were prone to clipping.

 

[Hlam]

Thanks Matthew
I have been in contact with Dirac. They are very responsive to emails.
I have never used REW before. When I get a chance I will start reading up on it. This forum seems to be the right place.
Not sure what you mean by, "wait for it to process that file". When I do a calibration I never move on to the next measurement position until that point is processed and turns green. Is this what you mean?
I use the laptop to play the Omnimic test disc. I learned the hard way to always make sure that everything is streamed through the DAPC.
I run the measurement levels so that the indicator just touches into the green. I found that when the level reaches -12 the tones are uncomfortable loud.

 

[Matthew J Poes]

Thanks Matthew
I have been in contact with Dirac. They are very responsive to emails.
I have never used REW before. When I get a chance I will start reading up on it. This forum seems to be the right place.
Not sure what you mean by, "wait for it to process that file". When I do a calibration I never move on to the next measurement position until that point is processed and turns green. Is this what you mean?
I use the laptop to play the Omnimic test disc. I learned the hard way to always make sure that everything is streamed through the DAPC.
I run the measurement levels so that the indicator just touches into the green. I found that when the level reaches -12 the tones are uncomfortable loud.

When you take measurements you don't have to wait for it to finish processing to take the next measurement. However you can't move on To generate a correction file until you do. I was just making the point that you can do a correction with a single measurement but need to wait a bit initially.

The reason the single measurement will be good is that it will be maximally smooth at a single measurement position. While trying to figure out what's going on, that can be helpful.

 

[Flak]

HLam sent me his project, we have to look at one channel at a time and follow AudiocRaver and Matthew's suggestions of taking measurements of a single channel in a single position.
Here the first image about frequency responses of Front Left channel before correction, all nine measurements in nine different positions are shown and it's clear that even if each of them is different the general behaviour is the one shown in the light blue curve (the average one)

while here the image about frequency responses of Front Left channel after correction:

Apparently there is a significant improvement, for completeness the impulse response should significantly improve as well:

Now I don't know the exact meaning of the previous graphs (there certainly is comb filtering if more than one channel are measured together) so HLam should look at one channel only at a time, Front Left in this case.
He could take a screen shot of Dirac graphs before and after correction as well as measure that same channel before and after correction with REW.
Mic position must strictly be the same and smoothing as close as possible (Dirac uses 1/8 of octave)
Dirac measurements in a single position cab be taken by clicking the Proceed button for all measurements after the first one.

Let's see what the outcome is...
Flavio

 

[Matthew J Poes]

HLam sent me his project, we have to look at one channel at a time and follow AudiocRaver and Matthew's suggestions of taking measurements of a single channel in a single position.
Here the first image about frequency responses of Front Left channel before correction, all nine measurements in nine different positions are shown and it's clear that even if each of them is different the general behaviour is the one shown in the light blue curve (the average one)

View attachment 6239

while here the image about frequency responses of Front Left channel after correction:

View attachment 6240

Apparently there is a significant improvement, for completeness the impulse response should significantly improve as well:

View attachment 6241

Now I don't know the exact meaning of the previous graphs (there certainly is comb filtering if more than one channel are measured together) so HLam should look at one channel only at a time, Front Left in this case.
He could take a screen shot of Dirac graphs before and after correction as well as measure that same channel before and after correction with REW.
Mic position must strictly be the same and smoothing as close as possible (Dirac uses 1/8 of octave)
Dirac measurements in a single position cab be taken by clicking the Proceed button for all measurements after the first one.

Let's see what the outcome is...
Flavio

Thanks for your support in this matter Flavio. Hopefully we can get to the bottom of it. There are so many variables in this right now that I need to have them reduced to feel confident in suggesting what might be going on. Because Dirac uniquely corrects the phase response of a speaker (which also better matches the phase between speakers) and because Audyssey does not do this, one theory I have is that this is simply the inter-channel correlation cancelation we would expect to see when two speakers are played. Precisely the needed phenomena for good imaging. If I can see a response measurement taken with REW (so that I can open it and examine it on my own laptop) then I may be able to see a change in the phase response that would suggest a substantial change was made by DIRAC. If we see just a single speaker measured and it looks much flatter, then I would feel confident this is what is going on (and is not a problem).

I've occasionally had results that came out of Dirac that looked great in DIRAC, but once measured with REW didn't look so good. In every case I could point to a particular problem on my part that caused this. One good example being that I attempted to create a wedge shaped response with DIRAC that amounted to 20db's of boost between 18khz and 20hz. Nearly all of the speakers actual response fell below the corrected curve and the amount of boost caused frequent clipping of Dirac (or at least this is my assumption of the cause of the problem). I redid this experiment moving the curve down in level so that there was no more than 6db's of boost relative to the average, and about 12db's of cut. When I took REW measurements with each correction filter and without, the test tones clipped the signal of one of the correction schemes (the first) and caused an ugly response and high distortion, for obvious reasons.

 

[Hlam]

Thanks Flavio and Matthew.
Flavio. Taking into consideration that you work for Dirac. Why are you working on Sunday?
I just had a look at REW for the first time.
This program may be over my head. It is definitely not plug and play like Omnimic. I will download and see where I get with it.
Thanks again guys.

 

[Flak]

Thanks Flavio and Matthew.
Flavio. Taking into consideration that you work for Dirac. Why are you working on Sunday?

Because I like this forum and my work

 

[Hlam]

I found REW to be a very frustrating program to set up.
I do think that I managed to get a set of frequency response measurements taken. See below.
These measurements are with no eq. If I follow REW instructions I can not make any measurements using Dirac. The REW instructions say to disconnect from the internet and Dirac's DAPC will not work without an internet connection.