Dirac Live Antics: A Different Goofball Thing With Every New Filter Attempt

[DavidMcRoy]

Should I just throw in the towel? I've tried to run setups of Dirac Live on a miniDSP Flex Eight a number of times, and it manages to ruin the sound in a different way each time. On my three attempts so far, It has (1) shifted the entire soundstage hard right, (2) rolled off the highs dramatically with lots of phasy effects in the lower midrange, and (3) flipped polarity between the left and right channels. (I am not kidding!) I have little faith I'll ever be able to get any satisfaction out of it. Is it just really buggy, inept, or both?

I'm going to put my IK Multimedia ARC Studio box back into signal path. It works flawlessly every time, additional A/D conversion required in my system be damned.

 

[Mort]

Use the curtain and avoid changing anything above ~300hz. That should help a lot.
What kind of mic are you using? Calibration file is there?

I found that it A) took a little more time than I expected to learn and B) I would still get a bad result 20% of the time.

 

[DavidMcRoy]

Use the curtain and avoid changing anything above ~300hz. That should help a lot.
What kind of mic are you using? Calibration file is there?

I found that it A) took a little more time than I expected to learn and B) I would still get a bad result 20% of the time.

I'm using a UMIK-1 with the correct calibration file. Full range in IK ARC4 works fine. Why shouldn't Dirac Live? And why would I be interested in reversed channel polarity below 300Hz?

 

[Mort]

I'm using a UMIK-1 with the correct calibration file. Full range in IK ARC4 works fine. Why shouldn't Dirac Live? And why would I be interested in reversed channel polarity below 300Hz?

Ok then don't.

 

[ppataki]

(1) shifted the entire soundstage hard right,

This has happened to me a lot. It is usually caused by a clock drift. I have replaced my UMIK-1 with an UMIK-2 (that has its own clock source so clock drift is no longer an issue) and the problem never happened again ever since

(2) rolled off the highs dramatically with lots of phasy effects in the lower midrange,

I had the same when I was measuring at 1 point only (the MLP), but if I do a proper 9-point measurement around the MLP it never came back

(3) flipped polarity between the left and right channels.

That never happened to me but I would take a strong bet that it is also clock drift related

Do you have a chance to try a UMIK-2 mic?

 

[DavidMcRoy]

This has happened to me a lot. It is usually caused by a clock drift. I have replaced my UMIK-1 with an UMIK-2 (that has its own clock source so clock drift is no longer an issue) and the problem never happened again ever since


I had the same when I was measuring at 1 point only (the MLP), but if I do a proper 9-point measurement around the MLP it never came back


That never happened to me but I would take a strong bet that it is also clock drift related

Do you have a chance to try a UMIK-2 mic?

Thanks! I'll get one to try out. (Never heard of this issue before.)

 

[Snickers-is]

This has happened to me a lot. It is usually caused by a clock drift. I have replaced my UMIK-1 with an UMIK-2 (that has its own clock source so clock drift is no longer an issue) and the problem never happened again ever since

That is interesting! I was not aware of this. But is this effectively solved in umik-2?

The reason why I ask is that since I (being a bit old school) am using a sound card with phantom voltage and condenser mics, and this may still be an issue if you're not using loop back.

So then I guess it will also be up to the software, because if it does not read the loop back, the available timing precision would be limited by the computer unless there are some very reliable ASIO setup that is also supported by the software.

But this may be besides the point. I think one of the most important functions of a room correction software is to identify what the stable room effects are, and filter out the unstable ones so there will be no attempt to correct for them as it would most likely degrade the sound quality. When a correction filter gives the kinds of variation described in OP, I would argue that there are unstable elements present in the correction process that are not filtered out.

I would argue that this is the main reason why room correction systems so often struggle to not degrade audio quality. Skipping all correction above 300 Hz will get rid of a lot of unstable elements, but overall timing errors may still be present.

 

[DavidMcRoy]

For what it's worth, when using the IK Multimedia ARC Studio hardware and their ARC4 Analysis software, have only used the dedicated ARC mic that comes with the processor, never the IMIK-1. Using their dedicated mic I've experienced no problems with ARC Studio. I haven't looked into whether the ARC mic is suitable for use with mniDSP.

 

[Mort]

It's not your microphone.

 

[Eric Natural]

Should I just throw in the towel? On my three attempts so far, It has (1) shifted the entire soundstage hard right . . .
I'm going to put my IK Multimedia ARC Studio box back

On the FlexDL ; I noticed an off-center reading and manually tweaked the levels back. None of your other weird behavior however. (wondering if the umic was hearing your air conditioning blowing or something)
Also using ARCv.x in my hobbyist studio DAW successfully.

 

[Oddball]

Should I just throw in the towel? I've tried to run setups of Dirac Live on a miniDSP Flex Eight a number of times, and it manages to ruin the sound in a different way each time. On my three attempts so far, It has (1) shifted the entire soundstage hard right, (2) rolled off the highs dramatically with lots of phasy effects in the lower midrange, and (3) flipped polarity between the left and right channels. (I am not kidding!) I have little faith I'll ever be able to get any satisfaction out of it. Is it just really buggy, inept, or both?

I'm going to put my IK Multimedia ARC Studio box back into signal path. It works flawlessly every time, additional A/D conversion required in my system be damned.

Every calibration with any mic will be a bit different than the other. There has been a discussion about UMIC1 vs UMIC2 lately, especially with reference to timing. I have no view on that discussion as never tried UMIC2, and likely not going to.

What you are describing though is unusual. If not failure in the process, some hardware seems to be malfunctioning. Many enjoy DIrac Live and its' older siblings with great success so it is not buggy, or inept. But then we all perceive things differently.

 

[DavidMcRoy]

It's not your microphone.

The only other possibilities you have mentioned are that Dirac Live may only be used successfully below 300Hz, and/or that more experience may be required on my part. As I mentioned, I've had zero issues getting ARC Studio to work correctly full range, and I'm sure I'm following the directions. Is there anything else that you think might be an issue?

 

[Mort]

As I mentioned, I've had zero issues getting ARC Studio to work correctly ful-rang

Many DSP's ignore above 300hz, including Genelec GLM. It usually does more harm than good. You just may be used to the sound the way the ARC studio adjusts above 300hz, but again, it's highly discouraged across all DSPs. Who knows how your hearing is calibrated. We get used to the sound we have over time and it sounds 'natural.'

 

[Mort]

You could post before and after FR's with REW and people here might spot something.

 

[DavidMcRoy]

On the FlexDL ; I noticed an off-center reading and manually tweaked the levels back. None of your other weird behavior however. (wondering if the umic was hearing your air conditioning blowing or something)
Also using ARCv.x in my hobbyist studio DAW successfully.

Spurious room noise can't be the issue. My room is below grade in a finished basement, and the noise floor is around +10dB or so.

 

[DavidMcRoy]

Many DSP's ignore above 300hz, including Genelec GLM. It usually does more harm than good. You just may be used to the sound the way the ARC studio adjusts above 300hz, but again, it's highly discouraged across all DSPs. Who knows how your hearing is calibrated. We get used to the sound we have over time and it sounds 'natural.'

I believe I'll try limiting the filter action to 300Hz or so. (My room actually "behaves badly" right AT around 300Hz.) Obviously, that isn't going to stop Dirac Live from inverting phase on one channel as it has done on one occasion.

 

[DavidMcRoy]

UPDATE: The UMIK-2 microphone arrived. I tried it, and no matter what I do I get a failed test due to "too much noise." There isn't any noise in the room, it's EXTREMELY quiet. I'm at a loss.

 

[Mort]

I was having that with Dirac and it was because I hadn't limited the channels to my actual speakers. If it tries to sweep a channel that has no speaker, it'll give the 'low signal to noise ratio' error.

 

[DavidMcRoy]

I was having that with Dirac and it was because I hadn't limited the channels to my actual speakers. If it tries to sweep a channel that has no speaker, it'll give the 'low signal to noise ratio' error.

That makes sense. In my case there are two 4-way stereo speakers, each with 7 drivers (a pair of woofers, a pair of upper bass woofers, a pair of mids, and a tweeter) requiring 4 channels for each speaker. See my profile pic.) That's a total of 8 channels for the stereo pair, so all channels are occupied, and all the drivers are working. And of course I wasn't getting a "too much noise" warning with my UMIK-1.

 

[EPS93]

FWIW I've experienced very similar issues with Dirac ART on my Denon 4800h, with soundstage shift left being the most common. My direct experience of the software has been that it is *extremely* buggy - I'm currently running an older version because the latest release broke on my Windows 11 laptop and no amount of reinstalls, manually fixing incorrect file locations present from install, resetting my AVR, etc. would convince Dirac that my Denon could actually be used. Even after that fix, it's not stable, and just now started throwing an error when I try to export filters (worked ~30 minutes ago, now failed twice).

Very interested to hear if anyone has fixed the soundstage shift issue - I got ART when it came out for Denon last year and after a few months of excellent performance, I've had this issue on at least half of my recent calibration attempts. I've used a umik-1 (with 90 degree calibration file) and a M23r running off a Topping e2x2 (with separate power), always use focused measurements and symmetrical locations, tried different measurement location patterns, etc., and nothing seems to consistently fix it. I recently came across this Dirac help article that suggests using the digital speaker trims during leveling to match "values between +/-0.2 dB to the quietest speaker level value" which I have not done (I try to keep levels fairly consistent, but not nearly that tight across all speakers) so I may try that tomorrow.