Dirac Live right-shifted soundstage due to added delay on right channel

[PHD]

Hi

It's been a few years since I had to run Dirac calibration on my NAD C658 (2.2 setup). Even after repeated attempts (re-measuring), I always end up with a right-shifted soundstage. If I change the balance to -4dB left, the soundstage returns to the center but still sounds off or out of phase.
I did some REW/Umik-1 measurements and it seems the issue isn't a level shift, but a time delay of about 0.7ms in the right channel as per the impulse response.

I've loaded the old Dirac filters (taken for different sets of speakers, and they don't have this issue but don't sound right bass-wise.

What could cause such behavior? I'm on the latest FW and Dirac version.

Thanks

 

[ban25]

Are your speakers close to a wall? Perhaps Dirac is picking up a wall reflection and confusing it as the primary impulse, hence the delay.

 

[pogo]

DL and BluOS are up to date?
Where was the mic connected?
Which OS was used for calibration?

 

[PHD]

Thanks. Speakers are 1 foot away from the walls and I don't get this issue when using Audyssey on my Denon AVR installed in the same room. Everything is up to date.

 

[ban25]

Thanks. Speakers are 1 foot away from the walls and I don't get this issue when using Audyssey on my Denon AVR installed in the same room. Everything is up to date.

Are those side-walls that are 1 foot away? In that case, you might want to try to move them a bit (if possible), or change your measurement position, or maybe try to apply some acoustic treatment to the walls. Or...just don't bother with Dirac and enjoy your setup as is.

 

[pogo]

Where was the mic connected?
Which OS was used for calibration?
Have you tried the NAD mic?

 

[pogo]

The UMIK-1 is not the best choice for DL: Link

 

[DVDdoug]

I don't know what's going-on (and I'm NOT a DIRAC expert) but there shouldn't be a delay unless you are sitting closer to one speaker than the other.

the issue isn't a level shift, but a time delay of about 0.7ms in the right channel as per the impulse response.

Sound travels at about 1-foot per millisecond so that's not much... About the distance between your ears, and you (and anyone else) won't always be sitting in the exact center.

 

[PHD]

Yes, I used the Umik-1 instead of the analog NAD mic. It's a Windows 11 laptop. I've repeated the 8 measurements today, but I used a laser distance meter to make sure the mic is placed symmetrically with respect to the MLP.

The results have improved. The soundstage is no longer shifted to the right, but something is still off, especially when mono dialog doesn't sound coherent and centered (phantom center).

Then I played Rpger Waters Amused to Death, which is Q-sound encoded. In the first tract (Ballad of Bill Hubbard) when Dirac is off or when playing the track using Denon X4800H with Audyssey (same room/speakers) in stereo, the god barking sound as if they're coming from the neighbor's yard next door. But once Dirac is on, the barking sounds like they're coming from the right speaker. So clearly, Dirac messes up the imaging and soundstage. I even tried limiting the curtain to only 300Hz, which didn't help.

I'll try to contact Dirac support. This is very disappointing. Audyssey does a much better job of maintaining the original soundstage.

 

[ban25]

Yes, I used the Umik-1 instead of the analog NAD mic. It's a Windows 11 laptop. I've repeated the 8 measurements today, but I used a laser distance meter to make sure the mic is placed symmetrically with respect to the MLP.

The results have improved. The soundstage is no longer shifted to the right, but something is still off, especially when mono dialog doesn't sound coherent and centered (phantom center).

Then I played Rpger Waters Amused to Death, which is Q-sound encoded. In the first tract (Ballad of Bill Hubbard) when Dirac is off or when playing the track using Denon X4800H with Audyssey (same room/speakers) in stereo, the god barking sound as if they're coming from the neighbor's yard next door. But once Dirac is on, the barking sounds like they're coming from the right speaker. So clearly, Dirac messes up the imaging and soundstage. I even tried limiting the curtain to only 300Hz, which didn't help.

I'll try to contact Dirac support. This is very disappointing. Audyssey does a much better job of maintaining the original soundstage.

Which calibration width are you using? Focused, wide, or widest? I would try using the widest one for a hopefully more forgiving calibration. Phantom center is a tough one, even with a single listener.

 

[pogo]

You must be 30dB above the noise level. It doesn't matter whether it is ambient noise or noise generated by the electronics. 75dB is not enough with a UMIK-1, for example!
All unnecessary Windows processes should also be stopped.
Dirac will also release a major update soon, as it still causes problems on some computer setups.
Btw., I only use a three-point measurement (focused).

 

[vicenzo_del_paris]

Yes, I used the Umik-1 instead of the analog NAD mic. It's a Windows 11 laptop. I've repeated the 8 measurements today, but I used a laser distance meter to make sure the mic is placed symmetrically with respect to the MLP.

The results have improved. The soundstage is no longer shifted to the right, but something is still off, especially when mono dialog doesn't sound coherent and centered (phantom center).

Then I played Rpger Waters Amused to Death, which is Q-sound encoded. In the first tract (Ballad of Bill Hubbard) when Dirac is off or when playing the track using Denon X4800H with Audyssey (same room/speakers) in stereo, the god barking sound as if they're coming from the neighbor's yard next door. But once Dirac is on, the barking sounds like they're coming from the right speaker. So clearly, Dirac messes up the imaging and soundstage. I even tried limiting the curtain to only 300Hz, which didn't help.

I'll try to contact Dirac support. This is very disappointing. Audyssey does a much better job of maintaining the original soundstage.

I have been using Dirac (Full range) for years with different speakers+subs in different rooms using minidsp products and umik-1.
Never got such issues : no left/right shift, no soundstage degradation.
Just tried the roger waters track and no difference with Dirac on/off for the barking dog.
The only (good) impacts on soundstage and imaging with Dirac are a more linear FR and reduced decay and reflections.

There must be something wrong during dirac calibration (mic issues, levels, room behaviour, ...) or with NAD integration.

 

[PHD]

I contacted Dirac support, but they keep referring me to a site I cannot access called ATLASSIAN. I get an error message saying that my email doesn't have access to Jira on dirac.atlassian.net.

@Flak

 

[pogo]

This beta could already solve your problem: Link

 

[Sammy135]

Hi

It's been a few years since I had to run Dirac calibration on my NAD C658 (2.2 setup). Even after repeated attempts (re-measuring), I always end up with a right-shifted soundstage. If I change the balance to -4dB left, the soundstage returns to the center but still sounds off or out of phase.
I did some REW/Umik-1 measurements and it seems the issue isn't a level shift, but a time delay of about 0.7ms in the right channel as per the impulse response.

I've loaded the old Dirac filters (taken for different sets of speakers, and they don't have this issue but don't sound right bass-wise.

What could cause such behavior? I'm on the latest FW and Dirac version.

Thanks

Hey man.

What a coincecdence I see this post. I litreally just solved this issue, and I was having this issue for over 2 months. I thought it was my speakers to be honest... I had just upgraded from kef ls50 meta to martin logan electrostats, and when I plugged my new speakers in, ran dirac, it started doing this...

the problem is speaker levels. Dirac might have got the delay right, but speaker level FORSURE it did not, hence why you have this problem.

measuring with a phone microphone or even calibrated SPL meter will not work the best. The pink noise generated by AVR is not enough bandwidth to use to level match. You have to do this by ear and by vocal heavy music playing through spotify, youtube, etc...

You say your balance was 4dB to correct, mine is 3.5db difference between left and right. Kind of insane.

So, because speaker levels were wrong, I think delay is also wrong. But you can't change delay once dirac is locked in. On my RZ50, delays are set to 3.5 ms, and I cant do anything about it. So, with a dB delta of 3.5 to 4 db, it can start to sound out of phase.

Like you, I solved the issue, but I am not sure what is causing it. Because visually looking at it, my left side doesnt look so different than the right side, that thered be a 3db difference.

 

[kyuu]

measuring with a phone microphone or even calibrated SPL meter will not work the best. The pink noise generated by AVR is not enough bandwidth to use to level match. You have to do this by ear and by vocal heavy music playing through spotify, youtube, etc...

On what basis do you claim that:

1) Pink noise does not have enough bandwidth to use for level matching, and

2) Your ears are better for level matching than a calibrated SPL meter?

 

[BJL]

Hi

It's been a few years since I had to run Dirac calibration on my NAD C658 (2.2 setup). Even after repeated attempts (re-measuring), I always end up with a right-shifted soundstage. If I change the balance to -4dB left, the soundstage returns to the center but still sounds off or out of phase.
I did some REW/Umik-1 measurements and it seems the issue isn't a level shift, but a time delay of about 0.7ms in the right channel as per the impulse response.

I've loaded the old Dirac filters (taken for different sets of speakers, and they don't have this issue but don't sound right bass-wise.

What could cause such behavior? I'm on the latest FW and Dirac version.

Thanks

There is a new (10/25/2024) update of Dirac to 3.11 which may resolve this. It appears in part to address distance alignment.

Dirac Live Beta 3.11 // 2024-10-25 | Software Changelog

Changes listed are relative to the latest public stable release.FeaturesDelay alignment update: The delay compensation for speaker distance alignment has previously been slightly confusing in that the shown delay values have not corresponded exactly to the measured acoustic delay. This has been...

helpdesk.dirac.com

 

[PHD]

Thanks.

I'll give it a try.

The NAD C658 doesn't allow user delay adjustments or speaker levels. It is a "black box" which can only be controlled directly by the loaded Dirac filters. It also lacks bass management, so I use miniDSP and MSO to align the subs using one sub output. Then I run Dirac calibration and after loading the filters, I use REW and try every possible cross-over setting in the NAD from 40Hz to 250Hz in 10Hz increments and compare the REW results to choose the best one. It's the only way to get it to sound right because DLBC isn't supported.

 

[PHD]

V3.11 has improved the issue but didn't fully resolve it. It may be an issue with the NAD C658 handling of delay compensation and Dirac v3.x

 

[pogo]

I would wait for the stable V3.11 and if this version does not solve the issue, perform a factory reset.