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

[klinemj]

Is this still the recommended way for running DIRAC? I've been out of it here for a while and lost track, but I'm about to re-run DIRAC and want to make sure I'm using the latest. (I have upgraded to Dirac 3.1.1 and the latest G3P FW.)

Mark

Put the Mic at 100%.
Start with Master extremely low, play pink noise on every channel to get the meters "settled" into a starting position, about two seconds for each channel.
Look for the channel with the lowest output and play the pink noise for that channel, in my system it's always the Surround Left, then move the Master slider up until you get the volume level you want for that channel, then stop playing the pink noise.
Now you can adjust all the other sliders, with no noise, until they all match the first channel you already set. When this is finished, go to the next step.
I shoot for about -26dB to -25dB when pink noise is playing for speakers, and a few dB higher for subs. This makes it loud enough but not dangerous to speakers or subs.
This takes me about a minute to do. It took me longer to type.

Run the sweeps for the one mic position, save the session before proceeding, then go to the next step which will bring up a nag screen "are you sure?" and yes, you're sure, so get to the next page and wait for the computing magic to happen, then go load the filter into the processor. I will always suggest doing the most simple process first to see what happens, then build on the results. Now you can compare 1-spot mic to 5-spot mic runs.

Also, you need to be certain that the mic can see every speaker directly. If a speaker is not clearly visible to the mic, then the mic is hearing reflections.

[leonski]

Is there a time window with those measures? The human ear / brain can't tell direct / reflected IF the time between is too short.

[marcl]

Sept 14, 2021 13:36:41 GMT -5 klinemj said:

Is this still the recommended way for running DIRAC? I've been out of it here for a while and lost track, but I'm about to re-run DIRAC and want to make sure I'm using the latest. (I have upgraded to Dirac 3.1.1 and the latest G3P FW.)

Mark

Put the Mic at 100%.
Start with Master extremely low, play pink noise on every channel to get the meters "settled" into a starting position, about two seconds for each channel.
Look for the channel with the lowest output and play the pink noise for that channel, in my system it's always the Surround Left, then move the Master slider up until you get the volume level you want for that channel, then stop playing the pink noise.
Now you can adjust all the other sliders, with no noise, until they all match the first channel you already set. When this is finished, go to the next step.
I shoot for about -26dB to -25dB when pink noise is playing for speakers, and a few dB higher for subs. This makes it loud enough but not dangerous to speakers or subs.
This takes me about a minute to do. It took me longer to type.

Run the sweeps for the one mic position, save the session before proceeding, then go to the next step which will bring up a nag screen "are you sure?" and yes, you're sure, so get to the next page and wait for the computing magic to happen, then go load the filter into the processor. I will always suggest doing the most simple process first to see what happens, then build on the results. Now you can compare 1-spot mic to 5-spot mic runs.

Also, you need to be certain that the mic can see every speaker directly. If a speaker is not clearly visible to the mic, then the mic is hearing reflections.

It has evolved and basically this is one way to do it, with the key thing getting all the channels around -25db or so.  You do want to get the quietest speaker to around -25db then you can pull the others down if necessary.

The part about doing one position first, I don't do that because Dirac needs several positions to work properly.  It will never sound its best with one position.  Five is minimum and nine is better.  The reason is Dirac has to compare measurements to determine which peaks and nulls are common and which are specific to different positions.  They don't want to do a literal average because they use different filter types and need to do a more complex analysis.

[marcl]

Sept 14, 2021 13:55:11 GMT -5 leonski said:

Is there a time window with those measures? The human ear / brain can't tell direct / reflected IF the time between is too short.

There certainly is a time window, but Dirac determines that.  We trust the PhD's over there to do measurements in a way that serves the purpose of creating the filters.  I have posted that my REW measurements and Dirac's measurements are essentially indistinguishable.

[klinemj]

Sept 14, 2021 13:56:06 GMT -5 marcl said:

Sept 14, 2021 13:36:41 GMT -5 klinemj said:

Is this still the recommended way for running DIRAC? I've been out of it here for a while and lost track, but I'm about to re-run DIRAC and want to make sure I'm using the latest. (I have upgraded to Dirac 3.1.1 and the latest G3P FW.)

Mark

It has evolved and basically this is one way to do it, with the key thing getting all the channels around -25db or so.  You do want to get the quietest speaker to around -25db then you can pull the others down if necessary.

The part about doing one position first, I don't do that because Dirac needs several positions to work properly.  It will never sound its best with one position.  Five is minimum and nine is better.  The reason is Dirac has to compare measurements to determine which peaks and nulls are common and which are specific to different positions.  They don't want to do a literal average because they use different filter types and need to do a more complex analysis.

Thanks - planning to do 9 positions. The level thing is the new thing vs. the last time I ran it.

Mark

[ttocs]

Sept 14, 2021 13:56:06 GMT -5 marcl said:

Sept 14, 2021 13:36:41 GMT -5 klinemj said:

Is this still the recommended way for running DIRAC? I've been out of it here for a while and lost track, but I'm about to re-run DIRAC and want to make sure I'm using the latest. (I have upgraded to Dirac 3.1.1 and the latest G3P FW.)

Mark

The part about doing one position first, I don't do that because Dirac needs several positions to work properly.  It will never sound its best with one position.  Five is minimum and nine is better.  The reason is Dirac has to compare measurements to determine which peaks and nulls are common and which are specific to different positions.  They don't want to do a literal average because they use different filter types and need to do a more complex analysis.

The recommending of doing a single mic position the first time is only so newbies can get one successful run completed.

[jbrunwa]

When I last ran Dirac Live my internet connection was briefly interrupted and the Dirac software errored. Fortunately I was able to save the session so I didn't have to take the measurements over again. It would be nice if the Dirac Live software did not require continuous internet connection to use - maybe check the license at the start of the session only. Also, it would be nice to be able to edit the filters without having to be connected to the AVR.

[AudioHTIT]

Sept 14, 2021 18:52:46 GMT -5 jbrunwa said:

When I last ran Dirac Live my internet connection was briefly interrupted and the Dirac software errored. Fortunately I was able to save the session so I didn't have to take the measurements over again. It would be nice if the Dirac Live software did not require continuous internet connection to use - maybe check the license at the start of the session only. Also, it would be nice to be able to edit the filters without having to be connected to the AVR.

I believe they do the heavy lifting of filter building on their home turf, so it’s more than checking the license.

[marcl]

Sept 14, 2021 21:48:38 GMT -5 AudioHTIT said:

Sept 14, 2021 18:52:46 GMT -5 jbrunwa said:

When I last ran Dirac Live my internet connection was briefly interrupted and the Dirac software errored. Fortunately I was able to save the session so I didn't have to take the measurements over again. It would be nice if the Dirac Live software did not require continuous internet connection to use - maybe check the license at the start of the session only. Also, it would be nice to be able to edit the filters without having to be connected to the AVR.

I believe they do the heavy lifting of filter building on their home turf, so it’s more than checking the license.

There is some controversy over this idea.   On one hand, Dirac clearly said years ago with version 1 that they did the filter processing on their servers and when making measurements I believe it even said something like "uploading measurements".  Back then there was also a reference to keeping the processing on their servers to preserve the secrecy of their filter calculations.  On the other hand, Keith has said they do the processing locally on your PC and that the thing about doing it on their servers was "marketing hype".  Now maybe it changed from V1 to V2/3?

On the "gripping hand" I observed a distinct difference in how measurement levels are set between late June and early July with out changing either Emotiva firmware or Dirac software versions (2.3, 3.0.14).  Nothing changed, and yet measurement levels were different than they had been for months, and have remained different since early July and with Dirac 3.1.1.  My process has been very well controlled for many months, so I know something else changed.

When I read descriptions of DLBC they said processing would be done mostly on their servers because it required a lot of processing power to get it done in a reasonable time.

I have asked Flavio in email and in a forum PM about all of this, and he didn't answer.

Bottom line - irrespective of where processing is done - they clearly are very conscious of preventing unauthorized use.  Theoretically, if they just authenticated the license once you could load filters on more than one processor, or something like that.  But for us, the license is tied to the MAC address of the device, not a license associated with the PC software.  It sure would be nice to be able to open Dirac and look at projects and change filter settings without having to connect.

So there we are ... clear as mud 

[KeithL]

Some of the early Dirac literature did give the impression that the processing was being done "on a server somewhere".

(I assume this was because some folks are impressed by the idea of massively complicated secret calculations being done on a supercomputer buried under a mountain somewhere...)

Also, in the past, some programs actually were copy protected by "holding back" a few small parts of the program, which were then loaded from a server when you ran it (so you never had a complete copy saved anywhere).

This did have some value in terms of making a program more difficult to hack or bootleg.

This is not done much lately - although today many so-called "web based applications" are actually loaded into your browser and run locally on your computer after that.

In general today servers are used mostly to store information - like your shopping cart and order information - because typical home computers have plenty of processing power to run software locally.

To be quite fair, assuming everything works as it should, it really doesn't matter.

In any case, as far as we know, the only information being exchanged with the servers, on the versions on both the XMC-1 and our new processors, is a license verification.

There are two parts to the Dirac Live software - the software itself and the "filter engine" on the processor.

The Dirac Live filters are NOT the same as ordinary PEQ settings... like the ones you get from REW... so you cannot "extract them and load them into an ordinary processor".

The Dirac Live filters can ONLY be used with a processor that includes the Dirac filter engine module in its firmware.

The version of Dirac Live that originally came with the XMC-1, and the full version you can purchase now for use with the XMC-1, are both linked to the person who purchases the license (user name and password).

And you can use a single copy of the software with more than one processor (but, since it is custom, it will only work on an Emotiva XMC-1).

With the latest version the license is tied to the processor instead.

And, when you run the program, it fetches the MAC address of the processor, and uses a secure hash of that to verify your license with the license servers.

(And you can download as many copies as you like, and install them on as many computers as you like, to use with your licensed processor.)

If you are really curious you can look at the network traffic easily enough with a packet sniffer like WireShark (since it is going over your network).

It is obviously up to Dirac what they allow you to do with the software when it is not connected to "a licensed unit".

(If you use other serious software you will realize that it is not all that unusual to have a program that is "only licensed for use with a single computer or device".)

However that is certainly a feature you could request directly from Dirac.

(It wouldn't be that difficult for them to allow you to edit and save projects while not connected but still prevent you from downloading the resulting filters to "an unlicensed processor".)

Sept 17, 2021 8:46:59 GMT -5 marcl said:

Sept 14, 2021 21:48:38 GMT -5 AudioHTIT said:

I believe they do the heavy lifting of filter building on their home turf, so it’s more than checking the license.

There is some controversy over this idea.   On one hand, Dirac clearly said years ago with version 1 that they did the filter processing on their servers and when making measurements I believe it even said something like "uploading measurements".  Back then there was also a reference to keeping the processing on their servers to preserve the secrecy of their filter calculations.  On the other hand, Keith has said they do the processing locally on your PC and that the thing about doing it on their servers was "marketing hype".  Now maybe it changed from V1 to V2/3?

On the "gripping hand" I observed a distinct difference in how measurement levels are set between late June and early July with out changing either Emotiva firmware or Dirac software versions (2.3, 3.0.14).  Nothing changed, and yet measurement levels were different than they had been for months, and have remained different since early July and with Dirac 3.1.1.  My process has been very well controlled for many months, so I know something else changed.

When I read descriptions of DLBC they said processing would be done mostly on their servers because it required a lot of processing power to get it done in a reasonable time.

I have asked Flavio in email and in a forum PM about all of this, and he didn't answer.

Bottom line - irrespective of where processing is done - they clearly are very conscious of preventing unauthorized use.  Theoretically, if they just authenticated the license once you could load filters on more than one processor, or something like that.  But for us, the license is tied to the MAC address of the device, not a license associated with the PC software.  It sure would be nice to be able to open Dirac and look at projects and change filter settings without having to connect.

So there we are ... clear as mud 

[jbrunwa]

Sept 17, 2021 10:42:46 GMT -5 KeithL said:

Some of the early Dirac literature did give the impression that the processing was being done "on a server somewhere".

(I assume this was because some folks are impressed by the idea of massively complicated secret calculations being done on a supercomputer buried under a mountain somewhere...)

Also, in the past, some programs actually were copy protected by "holding back" a few small parts of the program, which were then loaded from a server when you ran it (so you never had a complete copy saved anywhere).

This did have some value in terms of making a program more difficult to hack or bootleg.

This is not done much lately - although today many so-called "web based applications" are actually loaded into your browser and run locally on your computer after that.

In general today servers are used mostly to store information - like your shopping cart and order information - because typical home computers have plenty of processing power to run software locally.

To be quite fair, assuming everything works as it should, it really doesn't matter.

In any case, as far as we know, the only information being exchanged with the servers, on the versions on both the XMC-1 and our new processors, is a license verification.

There are two parts to the Dirac Live software - the software itself and the "filter engine" on the processor.

The Dirac Live filters are NOT the same as ordinary PEQ settings... like the ones you get from REW... so you cannot "extract them and load them into an ordinary processor".

The Dirac Live filters can ONLY be used with a processor that includes the Dirac filter engine module in its firmware.

The version of Dirac Live that originally came with the XMC-1, and the full version you can purchase now for use with the XMC-1, are both linked to the person who purchases the license (user name and password).

And you can use a single copy of the software with more than one processor (but, since it is custom, it will only work on an Emotiva XMC-1).

With the latest version the license is tied to the processor instead.

And, when you run the program, it fetches the MAC address of the processor, and uses a secure hash of that to verify your license with the license servers.

(And you can download as many copies as you like, and install them on as many computers as you like, to use with your licensed processor.)

If you are really curious you can look at the network traffic easily enough with a packet sniffer like WireShark (since it is going over your network).

It is obviously up to Dirac what they allow you to do with the software when it is not connected to "a licensed unit".

(If you use other serious software you will realize that it is not all that unusual to have a program that is "only licensed for use with a single computer or device".)

However that is certainly a feature you could request directly from Dirac.

(It wouldn't be that difficult for them to allow you to edit and save projects while not connected but still prevent you from downloading the resulting filters to "an unlicensed processor".)

Sept 17, 2021 8:46:59 GMT -5 marcl said:

There is some controversy over this idea.   On one hand, Dirac clearly said years ago with version 1 that they did the filter processing on their servers and when making measurements I believe it even said something like "uploading measurements".  Back then there was also a reference to keeping the processing on their servers to preserve the secrecy of their filter calculations.  On the other hand, Keith has said they do the processing locally on your PC and that the thing about doing it on their servers was "marketing hype".  Now maybe it changed from V1 to V2/3?

On the "gripping hand" I observed a distinct difference in how measurement levels are set between late June and early July with out changing either Emotiva firmware or Dirac software versions (2.3, 3.0.14).  Nothing changed, and yet measurement levels were different than they had been for months, and have remained different since early July and with Dirac 3.1.1.  My process has been very well controlled for many months, so I know something else changed.

When I read descriptions of DLBC they said processing would be done mostly on their servers because it required a lot of processing power to get it done in a reasonable time.

I have asked Flavio in email and in a forum PM about all of this, and he didn't answer.

Bottom line - irrespective of where processing is done - they clearly are very conscious of preventing unauthorized use.  Theoretically, if they just authenticated the license once you could load filters on more than one processor, or something like that.  But for us, the license is tied to the MAC address of the device, not a license associated with the PC software.  It sure would be nice to be able to open Dirac and look at projects and change filter settings without having to connect.

So there we are ... clear as mud 

Dirac tech support advised me that yes internet is required, to use a cell phone hotspot if home internet is not available and that traffic will be minimal and will not impact my data plan, so I doubt there’s much processing performed on their servers. One wonders what they are going to do now that they have licensed mass market manufacturers.

[leonski]

I have a utility on my Mac that allows me to watch the internet connection AND see how much data is moving each way.

Apple calls it 'Utility Monitor' and is in the Utilities folder...right on TOP.

OPINION?

Try that on the DIRAC software and take careful notes of how much data based on what you are doing.

MUST be a similar utility for Windows.....

[ttocs]

Sept 18, 2021 13:09:39 GMT -5 jbrunwa said:

Dirac tech support advised me that yes internet is required, to use a cell phone hotspot if home internet is not available and that traffic will be minimal and will not impact my data plan, so I doubt there’s much processing performed on their servers. One wonders what they are going to do now that they have licensed mass market manufacturers.

There are several key moments during a Dirac Run where network traffic is greatly increased.
When beginning a session and Dirac checks the license.
After Measurement, but before going to Filter Design.
Once in Filter Design and the predicted calcs are loaded on-screen.

From memory, which may be faulty, when I did this test, after I found out when the increase in network traffic would be occurring I recall disconnecting the network cable inbetween and all was well. But, just to be sure, I'll try this again sometime.

[AudioHTIT]

Sept 18, 2021 13:52:53 GMT -5 leonski said:

I have a utility on my Mac that allows me to watch the internet connection AND see how much data is moving each way.

Apple calls it 'Utility Monitor' and is in the Utilities folder...right on TOP.

OPINION?

Try that on the DIRAC software and take careful notes of how much data based on what you are doing.

MUST be a similar utility for Windows.....

I think you're referring to "Activity Monitor", and as often as I've used that App to monitor CPU / GPU usage, and see what Apps were hogging the processor, I'd never noticed the little tab that says "Network" and sure enough it will show you Sent & Received Bytes & Packets by App. I don't have Dirac on this Mac, but will try to remember to look at it next time on run it on my other mini. Thanks for that!

[leonski]

right you are......ACTIVITY Monitor. Top the the Utility folder found.....as it turns out......in 'finder'.....

[hsamwel]

Sept 3, 2021 14:39:28 GMT -5 zdoggz said:

Quick question (Dirac newbie) - if I want to do a dB boost on the low end similar to Harman curve, should I just apply the booster target curve on the sub channel or apply that curve on all channels?

Generally you should always have the same curve on all speakers..

But this less important for subs vs main speakers if you use them the normal way.
Normal meaning all small and one set of subs handling BM and LFE.

The thing you should think about is that about +/- 15hz (depending on slope) around
your crossover point the curves has to follow each other. Otherwise you’ll get a peak
or dip around it.
To get this perfect with different curves is a lot harder, that’s why it’s recommended
not to use different curves. There’s also other things like phase and stuff..

If you separate BM, LFE and have large speakers you practically have several bass sources
and everything Dirac does for your bass is basically out the window. With Dirac Bass Control
this will work fine as these are then calibrated together.
I guess Dirac Bass Control will handle the different curves as well because it handles all
crossovers.

[hsamwel]

Sept 2, 2021 6:40:34 GMT -5 swedishcaveman said:

Good DIRAC sound at last! My Imac have OSX 10.13, so I loaded DIRAC SW v3.0.0. IAW Dirac live changelog v3.0.0, they added unofficial support for MacOS 10.11, 10.12 and 10.13. Next I adjusted my sub´s high pass filter to 100hz, and low passed my Klipsch RF-63 tower speakers to 60hz, surround´s, rear´s and center to low pass 80hz. During volume setup before measurements I used values supplied by marcl and ttocs on this forum. -20db for the sub and -25db for all the rest. It´s a 7.1 set up no ATMOS. Mic gain 100% gave me -53db noise floor. I did an tight focus listening area measurement, using Storm Audio´s asymmetric mic placement pattern. Loaded a Harman6db curve to all channels.
The result are crazy good. Solid center stage, deep strong punchy bas and detailed surround effects moving all around my cave. I now realize that Emotiva Dirac user guide are the wrong way to go. As somebody wrote here, it is time for an update.

Emotiva already applies a 120hz LFE filter, with a 12dB slope if I remember correctly.
BM gets their own high pass filters with each crossover.. Don’t understand why you
need to use an external one?

It’s recommended to use the same crossover for atleast LCR.
You will NOT loose anything by setting you fronts to 80hz as the center. It will rather
improve sound quality..

-53dB is quite a high noise floor. Dirac recommends about 30dB above for best results.
But you’ll probably get by with 20-25dB above when measuring. As long as it doesn’t
complain about too much noise or clipping you’re fine.
Only the MLP is really really important to place correctly and dead center. The rest are
are mostly points that Dirac uses to see what happens around the MLP when doing its
thing.

From the Dirac Live Bass Control Guide. Note that Dirac doesn’t say every mic position
have to be in line of sight. Marcus (the Dirac guru on AVS) pretty much confirmed this as well.

Quote:
AustinJerry said:
Those are indeed good guidelines, although I don't understand why mention of maintaining a line-of-site to all speakers is not mentioned. I think this is an important guideline as well.
Marcus767 answered:
"maintaining a line-of-site to all speakers" is a good recommendation for setting up a listening room but it's irrelevant for a room correction system as it deals with the room as is not with the room as it should be.
End quote

Tips for a good measurement
There is no strict way of positioning the measurement points; however, there are some things that are worth to have in mind during the measurement procedure:

1. The basic principle is that any additional measurement improves the correction. However, depending on the acoustics of your room and equipment the benefit from more measurements may diminish faster. Therefore, note that you do not necessarily need to do all the measurements defined in the arrangement. However, we strongly recommend you to never do fewer than five measurements.

2. The measurement points should have a distance of at least 30 cm (12 inches) between each other.

3. Avoid making measurements in a too-small space. Even for the “Tightly focused” listening environment, it is important to spread out the microphone positions to a sphere of at least 1 metre in diameter. A too small space will result in over compensation and sound dry and dull.

4. Measure some points outside the listening area, e.g., for a sofa, it is recommended to do a few of the measurements 20-30 cm in any direction outside the couch.

5. Remember that you are measuring a volume rather than a surface and be certain that you take the measurements in different vertical positions as opposed to in a single horizontal plane.

6. Point the microphone upwards to the ceiling (90 degrees) when measuring to ensure that additional colouration from the microphone is similar for both the wall reflections and the direct wave from the speaker.

7. Remember that the positions specified in the arrangements act as a guide and you may deviate from them to put or de-crease emphasis on particular spaces.

[leonski]

2 questions?

First? What is 'microphone coloration' when dealing with a calibration microphone?

Second? If you repeat the measurments several times, what kind of repeatability do you expect in DB?

And finally? What would you expect the system repeatabilty limits to BE?

[hsamwel]

Sept 19, 2021 14:35:51 GMT -5 leonski said:

2 questions?

First? What is 'microphone coloration' when dealing with a calibration microphone?

Second? If you repeat the measurments several times, what kind of repeatability do you expect in DB?

And finally? What would you expect the system repeatabilty limits to BE?

If the sound hits the microphone at the wrong angle it was calibrated to you’ll get different
results. That’s why it’s so important to place the mic straight up for each measure point.
Also all microphones used for home usage, like UMIK-1 or UMIK-2 are not industry spec.
You can’t expect them to measure as correct as a $5000 microphone.

Dirac does as good as it can with the gear you use with it, and also how the gear is
used. Sometimes it gets it wrong and you’ll have to redo the measurement.
If users follow Diracs recommendations they will get good results more often than
not..

100% repeatability is impossible with the kind of gear we’re using.
Also most of us don’t have robots or lasers to place the mic at exact positions.
We also measure in homes. This means something may have been moved or
added between measurements.

I have done about ten complete measurements and all but one sounded great.
Although all of them looked about the same in the frequency graphs.

I’d say Dirac can repeat its measument pretty good as long as the speaker or
room doesn’t do anything really odd. Even with the cheap gear we’re using.
Just get the first mic position dead center and pointing straight up.

[p4t]

A quick question. Does dirac live using peq for fixing freq response? If they do, maximum how many number of peq does dirac capable of? Because if we do our on peq on xmc-2, only capable of 10 or 11 peq maximum.