The Official Dirac Live Thread: Start w/1,2,3 🙋‍♂️

[marcl]

So what is different?  Anyone?  Anyone?  Bueller?

Two different calibrations with Dirac 3.11.   I did the first, uninstalled and reverted to 3.10.10, then uninstalled and reinstalled 3.11 ... and did a new set of measurements.  Look at the difference between the large L/R fronts and the Bass Management in the small speakers.  Note that Bass Management from small speakers goes to the large front L/R



There is one small difference that does not affect the Bass Management difference.  In Cal1 the L/R target is flat.  In Cal2 the L/R target has a slight shelf below 100Hz.  But ... I had applied the shelf to Cal1 and the Bass Management levels rose proportionately.  

[okjazz]

Nov 14, 2024 10:10:23 GMT -5 marcl said:

So what is different?  Anyone?  Anyone?  Bueller?

Two different calibrations with Dirac 3.11.   I did the first, uninstalled and reverted to 3.10.10, then uninstalled and reinstalled 3.11 ... and did a new set of measurements.  Look at the difference between the large L/R fronts and the Bass Management in the small speakers.  Note that Bass Management from small speakers goes to the large front L/R

View AttachmentView Attachment

There is one small difference that does not affect the Bass Management difference.  In Cal1 the L/R target is flat.  In Cal2 the L/R target has a slight shelf below 100Hz.  But ... I had applied the shelf to Cal1 and the Bass Management levels rose proportionately.  

Have done any listening test of the two calibrations? If so, which one did you like the most? The second calibration seems to have flattered some sound frequencies in the base region, filled up some nulls in the 500hz region, and smoothed some roll-offs in the highest frequencies region.

[marcl]

Nov 14, 2024 12:09:33 GMT -5 okjazz said:

Nov 14, 2024 10:10:23 GMT -5 marcl said:

So what is different?  Anyone?  Anyone?  Bueller?

Two different calibrations with Dirac 3.11.   I did the first, uninstalled and reverted to 3.10.10, then uninstalled and reinstalled 3.11 ... and did a new set of measurements.  Look at the difference between the large L/R fronts and the Bass Management in the small speakers.  Note that Bass Management from small speakers goes to the large front L/R

View AttachmentView Attachment

There is one small difference that does not affect the Bass Management difference.  In Cal1 the L/R target is flat.  In Cal2 the L/R target has a slight shelf below 100Hz.  But ... I had applied the shelf to Cal1 and the Bass Management levels rose proportionately.  

Have done any listening test of the two calibrations? If so, which one did you like the most? The second calibration seems to have flattered some sound frequencies in the base region, filled up some nulls in the 500hz region, and smoothed some roll-offs in the highest frequencies region.

Yes, and the second calibration sounds better and is the more "correct" calibration.   It did do a bit better job in the higher frequencies too.

This was a bit of a trick question.  The difference between the two calibrations is not Dirac doing a better job of correcting the bass frequencies.  It's related to the legendary Bass Management bug.  This is the third time I've documented an effect when going to a Dirac major release where the first measurement and calibration that I do results in Bass Management being broken.  The two should not be connected, but they are.  The symptom is either as you see here, Bass Management of small speakers is elevated 7-8db higher than it should be by the processor (not Dirac);  or the other symptom is that Bass Management does not work at all and bass from small speakers disappears.  

The difference between the two calibrations is that after having done the calibration with 3.11 ...  I uninstalled 3.11, installed 3.10.10, tried to load measurements from 3.11 (which didn't work, because you can't go back from a major release), uninstalled 3.10.10, reinstalled 3.11, took a NEW set of measurements with exactly the same setup as previous and similar flat target curves .... and the Bass Management bug went away.

I'm posting this again (hasn't happened in a couple years) in case anyone else runs into this strange phenomenon ... so's we don't chase our tail like I did past few days. 

[webmst007]

Hi Marc - once again - thank you for all your practical experiments.
There's really no substitute for real world experiences. 😁💯👍

One factor in these readings I think should also be noted is the inconsistency inherent in measurements of this nature.

A 8kHz signal has only around a 4cm wavelength and a 10kHz sound is only 3cm.
So in theory it only needs a slight change in either recording device (mic placement) or surrounding environment to potentially alter results in a reasonably sensitive calibration system.

Maybe a quick set of Dirac readings at say 5mm changes in mic location would test that theory.
I'm not set up right now to do this - but it might be interesting to see what the effects are.

[marcl]

Nov 14, 2024 17:12:09 GMT -5 webmst007 said:

Hi Marc - once again - thank you for all your practical experiments.
There's really no substitute for real world experiences. 😁💯👍

One factor in these readings I think should also be noted is the inconsistency inherent in measurements of this nature.

A 8kHz signal has only around a 4cm wavelength and a 10kHz sound is only 3cm.
So in theory it only needs a slight change in either recording device (mic placement) or surrounding environment to potentially alter results in a reasonably sensitive calibration system.

Maybe a quick set of Dirac readings at say 5mm changes in mic location would test that theory.
I'm not set up right now to do this - but it might be interesting to see what the effects are.

5mm?  Dude ... I'm only interested in like the variation from one side to the other of my old bald head 

I actually measured after a Dirac calibration once, in an area about 6" around the central mic position and there was VERY little difference.  In my room if you go a couple feet either way the bass changes.  But otherwise, yeah maybe some changed due to comb filtering but if you sit still it's stable.  If you apply psychoacoustic smoothing to the measurements it approximates how our brains low pass the whole thing.

I think it also has to do with how live the room is.  My room has a lot of diffusion, not a lot of hard reflection, and  in the corners and some near the ceiling a lots of bass absorption.  The result is that it's very easy to localize sound sources, even down to 30Hz!  In a room that is much more live - as most rooms are - with decay times over 500ms, I think the localization of sound becomes more difficult and probably also the perception of tonality differences across a few feet from the MLP ... it's all bouncy chaos.

[lgjr]

Hello,

CSL umik1 or umik2 for calibration

[ttocs]

Nov 15, 2024 9:17:56 GMT -5 lgjr said:

Hello,

CSL umik1 or umik2 for calibration

Get the UMIK-2, and then request from miniDSP the app called "miniDSP-MIK" (you'll need a miniDSP.com login to do this). The app will give you control over the Sensitivity of the mic without the need to open the mic like what must be done with the UMIK-1.

The UMIK-2 has a different sized capsule, has slightly better audio specs, but the big benefit is being able to control the Sensitivity without physically opening up the mic and changing DIP switches.

I've got many mics, as seen in my avatar, and my CSL mic isn't any better than any of the other UMIK mics. CSL just calibrates to a different reference microphone, that's it.

Let me say this again, and it won't be the last time. ALL miniDSP UMIK microphones are INDIVIDUALLY CALIBRATED, and have been for about five years. This used to be a reason for getting a CSL calibrated UMIK back when the miniDSP purchased mics were batch calibrated. But miniDSP stopped using a batch cal and went to individual calibrations for every mic.

[webmst007]

Interesting Marc - thanks.
I agree - that test might play out differently in a "live" room.
I've only run Dirac on my umik 2 at the standard sensitivity setting.

[marcl]

Nov 15, 2024 17:36:40 GMT -5 webmst007 said:

Interesting Marc - thanks.
I agree - that test might play out differently in a "live" room.
I've only run Dirac on my umik 2 at the standard sensitivity setting.

There is a test I did a couple years ago that was pretty conclusive and interesting.  After a lot of speculation around how Dirac measurements prior to calibration, Dirac predicted response, and REW measurements after calibration could vary based on the sample length, timing windows, etc. ... I did a single point Dirac calibration and without moving the mic I did an REW measurement.  The REW frequency and impulse responses matched the Dirac predicted responses EXACTLY! 

Now in practice there's good reason to do at least 5 or more Dirac measurements spaced at least a few inches, so Dirac can parse out minimum phase vs non-minimum phase anomalies and devise an optimal filter scheme that does as much good as possible, with no harm.  Inevitably any measurement you take with REW won't match the predicted AS exactly but pretty close.  And so REW measurements within the envelope of the Dirac measurements should also be pretty close.

[LCSeminole]

[marcl]

Nov 16, 2024 16:33:52 GMT -5 LCSeminole said:

I watched this and there's a lot of good information here.  One of the most useful things is suggestions around using a good quality boom stand and having some process that makes the measurements go faster ... AND makes them more consistent from one calibration to the next.  I know many people dread doing a calibration because they fumble around trying to get the mic in position.  I'm lucky my desk is behind the couch, and with just two steps I can reposition my boom without moving the base and reach all the positions.  I can get six positions measured in less than 20min.

As for the number of measurements, I can see how each product may suggest a different number of measurements based on what they actually DO with the measurements ... how many do an arithmetic average vs something more sophisticated like Dirac looking at minimum phase vs non-minimum phase deviations from target?  That's all fine ... don't take too many, and don't take them too close together ... and never within 3ft of a room boundary.

Patterns?  I would really like to know WHY!  I have never heard an explanation from anyone as to why you should take measurements on one level or some higher or lower.  I see Theo does some only +/- an inch or two?  Like half wavelength of a 4KHz wave?  And you're probably not even correcting that high?  It will do no harm ... may not matter.

In the end this is very useful and following these recommendations will help most people.   The thing he doesn't mention is doing a post-calibration measurement with REW.  THAT closes the loop.  THAT lets you corelate what you see in the prediction to what actually happens in the room ... and whether one method or another makes a difference in your room.  And it also gives you a way to tweak a target curve without taking new measurements ... and see if the response can be improved.

[lgjr]

Nov 15, 2024 10:20:24 GMT -5 ttocs said:

Nov 15, 2024 9:17:56 GMT -5 lgjr said:

Hello,

CSL umik1 or umik2 for calibration

Get the UMIK-2, and then request from miniDSP the app called "miniDSP-MIK" (you'll need a miniDSP.com login to do this). The app will give you control over the Sensitivity of the mic without the need to open the mic like what must be done with the UMIK-1.

The UMIK-2 has a different sized capsule, has slightly better audio specs, but the big benefit is being able to control the Sensitivity without physically opening up the mic and changing DIP switches.

I've got many mics, as seen in my avatar, and my CSL mic isn't any better than any of the other UMIK mics. CSL just calibrates to a different reference microphone, that's it.

Let me say this again, and it won't be the last time. ALL miniDSP UMIK microphones are INDIVIDUALLY CALIBRATED, and have been for about five years. This used to be a reason for getting a CSL calibrated UMIK back when the miniDSP purchased mics were batch calibrated. But miniDSP stopped using a batch cal and went to individual calibrations for every mic.

RMC1,ATI SIGNATURE 6003,ATI AT5228 8CH AMP,LG OLED CX77, OPPO 203, PROJECT XPRESSION CARBON,LUMAGEN PRO, LEGACY FOCUS SE Front Main,LEGACY MARQUIS Center,LEGACY PHANTOMS Surrounds,LEGACY PHANTOMS Rear Surrounds,RYTHMIK FV25HP(2) & XTZ Sub3(2)

Thx

[danielb]

Is there a way to convert, or import a 5.1 measurement into a 2.0 measurement?
In my environment, the room and the electronics are the same for 5.1 or 2.0 preset. This would save me an additional series of measurements for the same speakers - which I have already measured.

Thanks again for any tips, if there are any.

[marcl]

Nov 18, 2024 10:12:05 GMT -5 danielb said:

Is there a way to convert, or import a 5.1 measurement into a 2.0 measurement?
In my environment, the room and the electronics are the same for 5.1 or 2.0 preset. This would save me an additional series of measurements for the same speakers - which I have already measured.

Thanks again for any tips, if there are any.

When you set the audio mode to 5.1 or stereo it uses the same Dirac filter.  So you don't have to do anything.

If you want to have two presets - possibly with a different target curve to the filters for 2.0 vs 5.1 - you can reload the filter project with the speakers set to 5.1, change the targets for the L/R and save to a new filter name, then in the Preset just disable the speakers you aren't using.  The Dirac filter will work just the same for your L/R.

[marcl]

A general answer to the previous question ...

• When you do a Dirac calibration, do it with the most speakers you will ever use, and save the Measurements as a Project before going to Filter Design.  That way you can reload the Measurements and add or delete if you want.  But you always have to have the Preset set to the same speaker configuration you used originally for the measurements.
• You can save 3 Dirac filter sets per Preset.  They can all be made from the same Measurements, but with different Target Curves, Curtains, etc. ... even if you don't intend to have all the speakers in use.
• Once you select which Dirac filter set you want to use in the processor it will work for whatever audio mode or speaker configuration is selected.  For example:
  - You can select Stereo, Dolby Surround, All Stereo .... whatever speakers are playing, they will use whatever Dirac filter is selected.

         -  You can disable some speakers in a Preset ... Dirac will use whatever filter set is selected for whatever speakers are enabled.

Sometimes I play a stereo recording in Stereo mode ... sometimes I play it with the Dolby Surround upmixer.  Same Dirac filters are used and the appropriate speakers play.

One thing that I have done is have Music and Video Presets with identical Dirac filters loaded to both.  In the Video Preset Center Sub=LFE so LFE goes to the Center Sub and Bass Management goes to the Large Fronts.  In the Music Preset it's identical from Dirac's perspective, but I say Center Sub=None .... so it plays Bass Management and LFE through the Large Fronts.  The only thing I have to remember is if I want to load a new Dirac filter set into the Music Preset, I have to turn the Center Sub back to LFE or else Dirac will not let me load a previous Project.

[lgjr]

Nov 15, 2024 10:20:24 GMT -5 ttocs said:

Nov 15, 2024 9:17:56 GMT -5 lgjr said:

Hello,

CSL umik1 or umik2 for calibration

Get the UMIK-2, and then request from miniDSP the app called "miniDSP-MIK" (you'll need a miniDSP.com login to do this). The app will give you control over the Sensitivity of the mic without the need to open the mic like what must be done with the UMIK-1.

The UMIK-2 has a different sized capsule, has slightly better audio specs, but the big benefit is being able to control the Sensitivity without physically opening up the mic and changing DIP switches.

I've got many mics, as seen in my avatar, and my CSL mic isn't any better than any of the other UMIK mics. CSL just calibrates to a different reference microphone, that's it.

Let me say this again, and it won't be the last time. ALL miniDSP UMIK microphones are INDIVIDUALLY CALIBRATED, and have been for about five years. This used to be a reason for getting a CSL calibrated UMIK back when the miniDSP purchased mics were batch calibrated. But miniDSP stopped using a batch cal and went to individual calibrations for every mic.

Once I get the mik app, what should the sensitivity be set to?

[ttocs]

Nov 19, 2024 7:09:37 GMT -5 lgjr said:

Nov 15, 2024 10:20:24 GMT -5 ttocs said:

Get the UMIK-2, and then request from miniDSP the app called "miniDSP-MIK" (you'll need a miniDSP.com login to do this). The app will give you control over the Sensitivity of the mic without the need to open the mic like what must be done with the UMIK-1.

The UMIK-2 has a different sized capsule, has slightly better audio specs, but the big benefit is being able to control the Sensitivity without physically opening up the mic and changing DIP switches.

I've got many mics, as seen in my avatar, and my CSL mic isn't any better than any of the other UMIK mics. CSL just calibrates to a different reference microphone, that's it.

Let me say this again, and it won't be the last time. ALL miniDSP UMIK microphones are INDIVIDUALLY CALIBRATED, and have been for about five years. This used to be a reason for getting a CSL calibrated UMIK back when the miniDSP purchased mics were batch calibrated. But miniDSP stopped using a batch cal and went to individual calibrations for every mic.

Once I get the mik app, what should the sensitivity be set to?

A test is needed to find out if your computer sees the UMIK-2 with a USBC connector on it as a mic with 18dB gain or ignores that setting. So, . . . . 

Connect your EMM-1 mic and the UMIK-2 mic and proceed to Dirac Volume Calibration.
Set the Mic Gain at 100%,
Set the Master Output to a low setting at first,
Play the pink noise on the first channel and raise the Master Output a little to get a comfortable sound level, then play the pink noise on a couple more channels,
Make a screenshot of Volume Calibration.
Now back out of Volume Calibration and go back to the start and select the UMIK-2 mic and proceed to Volume Calibration.
Using the screenshot, make the settings for Master Output and Mic Gain the same as before,
Play the pink noise on the same few channels as before,
Make another screenshot.

Compare both screenshots noting the level of the speaker channels. How different are they? Are they just a few dB different, or more than 10dB different?

Please report back with the info from above. I've done lots of testing on this and can tell you exactly which setting to use with the miniDSP-MIK app once you tell me what you discover. It might help to post the images to make things easier as I can interpret the differences very quickly.

[ttocs]

Nov 19, 2024 7:09:37 GMT -5 lgjr said:

Once I get the mik app, what should the sensitivity be set to?

Here are some examples using my system.

Here is Volume Calibration using my EMM-1 mic and the cal file as supplied by Emotiva.



Here is the same Master Output and Mic Gain settings in Volume Calibration, but with a UMIK-2 set for 18dB gain using the miniDSP-MIK app.



And lastly, here is the UMIK-2 with a 32dB gain setting using the miniDSP-MIK app.


Note the changes in Speaker Levels with each microphone.

Firstly, I should point out that I didn't make up using 32dB gain out of nowhere, I did tests using a UMIK-1 mic with the USBA connection set for 18dB gain. I then used the miniDSP-MIK app and increased the analog gain until it matched what Dirac would see when using the UMIK-1/USBA mic. All the new UMIK mics now come with USBC connections on them. These don't act the same as the older mics with USBA connections on them. For some reason, and with Dirac in particular, the Sensitivity setting inside the mic is ignored. I won't get into the offsets involved, but just know that what I think is happening is the Sensitivity setting in the USBC mics is completely ignored so, in the case of the UMIK-2, we need to use the miniDSP-MIK app to change the ANALOG gain of the mic. THIS IS NOT THE SAME AS USING THE MIC GAIN SLIDER IN DIRAC! The Mic Gain slider in Dirac uses DIGITAL GAIN once it goes above 100%. Digital Gain = BAD. Analog Gain = GOOD.

The EMM-1 mic has a different amount of gain than the UMIK mics that have 18dB gain settings. The EMM-1 has about 12dB of gain, or pretty close to that.

So the simple answer to your question is either 18dB or 32dB of analog gain as seen in the miniDSP-MIK app. It all depends on your computer. There was a patch supplied by Microsoft last year that kinda fixed this issue with USBC mics as used with Dirac. I don't recall any particulars about the patch as I use Macs for almost everything. But when using a UMIK-2 it really doesn't matter since we can change the gain quickly and easily.

Important: Please note that the 32dB gain setting for usage with Dirac, if needed, won't work when using REW which recognizes the USBC mics as intended so you'll need to use the 18dB setting with REW.

[marcl]

Since seeing a video (which I can't seem to find) where a guy spent days trying various Dirac measurement patterns until he found one that seemed to be best for a single listener ... I've been using this "Six-Pack" or "Domino" pattern ... roughly depicted here ...


All the measurements are on the same plane 8" apart, with the critical MLP measurement #1 in the middle of the rear row.  It works great!

So after watching the video someone posted recently where Theo from Audioholics goes through measurement patterns for Dirac, Audyssey and ARC ... and after recently watching some videos with Carl Tatz who designs mixing rooms with his "Phantom Center" method ... I got to thinking that maybe I'd try doing some high and low measurements along with measuring the MLP a little more behind my actual ears.

Tatz puts the focal point of his room timing 20" behind the MLP.  I wasn't going to go to that extreme and it really doesn't work with multichannel unless you have individual control of each speaker's  distance, which we don't with Dirac.

So okay I tried to do a multiple exposure to show this new pattern relative to my actual ears... black cap in place so as not to glare off my old bald head


I recline the couch all the way when I measure, and pile on a blanket to absorb close reflections from the leather.  So you see the MLP measurement is maybe a foot behind where my ears are when I adjust the couch for actual listening.  Then I went two high in the rear on either side.  Then 8" forward with two low L/R, and the center higher than the MLP was.

I'm listening now.  A lot of stuff sounds really great and that could be that Dirac 3.11 is doing something new too.  The imaging and soundstage is different and in many cases especially Atmos and 5.1 music it really seems to be better.  Feels more forward and three dimensional.  I feel like some stereo tracks sound different aside from imaging and soundstage.  Have to listen some more and give it time.

p.s.  Easier to imagine taking 7 photos and blending them in Photoshop, than it turned out to be.  But I hope you get the idea from this composite.  I don't sit in the seat when I measure.

[geebo]

Nov 20, 2024 12:27:51 GMT -5 marcl said:

Since seeing a video (which I can't seem to find) where a guy spent days trying various Dirac measurement patterns until he found one that seemed to be best for a single listener ... I've been using this "Six-Pack" or "Domino" pattern ... roughly depicted here ...
View Attachment

All the measurements are on the same plane 8" apart, with the critical MLP measurement #1 in the middle of the rear row.  It works great!

So after watching the video someone posted recently where Theo from Audioholics goes through measurement patterns for Dirac, Audyssey and ARC ... and after recently watching some videos with Carl Tatz who designs mixing rooms with his "Phantom Center" method ... I got to thinking that maybe I'd try doing some high and low measurements along with measuring the MLP a little more behind my actual ears.

Tatz puts the focal point of his room timing 20" behind the MLP.  I wasn't going to go to that extreme and it really doesn't work with multichannel unless you have individual control of each speaker's  distance, which we don't with Dirac.

So okay I tried to do a multiple exposure to show this new pattern relative to my actual ears... black cap in place so as not to glare off my old bald head
View Attachment

I recline the couch all the way when I measure, and pile on a blanket to absorb close reflections from the leather.  So you see the MLP measurement is maybe a foot behind where my ears are when I adjust the couch for actual listening.  Then I went two high in the rear on either side.  Then 8" forward with two low L/R, and the center higher than the MLP was.

I'm listening now.  A lot of stuff sounds really great and that could be that Dirac 3.11 is doing something new too.  The imaging and soundstage is different and in many cases especially Atmos and 5.1 music it really seems to be better.  Feels more forward and three dimensional.  I feel like some stereo tracks sound different aside from imaging and soundstage.  Have to listen some more and give it time.

p.s.  Easier to imagine taking 7 photos and blending them in Photoshop, than it turned out to be.  But I hope you get the idea from this composite.  I don't sit in the seat when I measure.

This video?