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

[ttocs]

Jul 26, 2020 16:05:17 GMT -5 megash0n said:

When looking at the impulse response tabs, most all of the speakers moved out to 20-30 ms instead of something closer to zero. If that makes sense.

Mine moved to 16ms. I think it's taking the longest delay, adding a little, and aligning everything to the new number. Just guessing. Kind of like what it does to overall levels, level to the weakest link.

edit: FWIW I'm mostly just doing tests with a single mic location. It's really quick this way to get a handle on what's going on.

[davidl81]

Ok, so I’m on a flight right now I I tried to read through all of this info. Here is my plan and let me know if I’ve got this down.
1) Install FW 2.0
2) Restore RMC-1
3) Set up active speakers, but leave levels and distance at 0
4) Set sub gain at “50%”?
5) Run Dirac and take 7000 measurements (I’m drinking wine on this flight so cut me some slack)
6) My biggest question. Use Dirac House curve or use one of the Harmon curves. I don’t know what is the best option on this one.
7) Post on this forum and complain about how something does not work right.

Let me know if this all makes sense.

[ttocs]

Jul 26, 2020 16:50:49 GMT -5 megash0n said:

Jul 26, 2020 16:18:33 GMT -5 ttocs said:

Mine moved to 16ms. I think it's taking the longest delay, adding a little, and aligning everything to the new number. Just guessing. Kind of like what it does to overall levels, level to the weakest link.

edit: FWIW I'm mostly just doing tests with a single mic location. It's really quick this way to get a handle on what's going on.

Are you not hearing anything that resembles what I am? I spent a lot of today wondering if I just prefer bad sound and can't appreciate how it is supposed to sound.

No. Things seem pretty clear. And the system cohesiveness is pretty strong.

I really haven't been able to do proper calibration sessions since finishing some construction in the house last week and having just finished properly mounting the Tops, plus swapping around which amps do what. So everything I've done so far is all just tests.

Plus, since the filters I made before a few days ago do not include the Tops, I can't even look at those in Dirac unless I change the Speaker Settings so the speaker layout matches the filter. This is one great reason to always measure all the speakers even if you're not going to use them in a particular setup.

[megash0n]

As someone suggested via PM, I'm going to actually flatten the curve and make sure nothing is boosted. I may have the bottom end raised a hair, but maybe this dead stuff is due to using the curves that lower the top end. Dialogue doesn't seemed to be much different. Just more precise and discrete. I don't seem to have good sub integration and my highs seem dead.

[ttocs]

Jul 26, 2020 17:13:43 GMT -5 davidl81 said:

Ok, so I’m on a flight right now I I tried to read through all of this info. Here is my plan and let me know if I’ve got this down.
1) Install FW 2.0
2) Restore RMC-1
3) Set up active speakers, but leave levels and distance at 0
4) Set sub gain at “50%”?
5) Run Dirac and take 7000 measurements (I’m drinking wine on this flight so cut me some slack)
6) My biggest question. Use Dirac House curve or use one of the Harmon curves. I don’t know what is the best option on this one.
7) Post on this forum and complain about how something does not work right.

Let me know if this all makes sense.

KISS Start really basic. I started with the 9 point measurement because I thought I had to, but on the second attempt I found that I could do just a single mic point. This helped with getting to know the process and learn the app.

Have another glass.

[Lonnie]

Jul 26, 2020 15:52:39 GMT -5 megash0n said:

Jul 26, 2020 15:06:37 GMT -5 Lonnie said:

I’m trying to get a handle on your system to better understand. So you are going out of the processor to a Mini DSP and then to the amps? Is that correct?

Thanks,

Lonnie

Thanks Lonnie.  I have minimized this issue some by switching the miniDSP input jumper to 2V instead of 4V which doesn't make a ton of sense to me, but it gave me roughly 5dB more sound by doing so.   To answer your question, my RMC-1 uses Emotiva XLR cables to a Gen 2 XPA-5 and a Gen 3 XPA four.  All speakers are Emotiva. The Center Sub outputs thru a handmade XLR cable that feeds a MiniDSP 2x4 Balanced on input 1.  The MiniDSP connects output 1 & 2 to a Behringer EP4000 amp via XLR cables.  The EP4000 is in stereo mode supposedly outputing 950 watts RMS (each) to a pair of sealed 12" TC Sounds LMS-R subs.  These subs are rate at 1000 watts RMS each.  I have used MSO to combine the two subs in a very acceptable way using a "flat or no curve".   After all my efforts on this, and all levels set to zero, the center sub is still roughly 5dB less than the mains using the test tones on the processor.
I reran Dirac quickly selecting only (3) measurement points at my MLP because I wanted to see if bringing the subs up helped. ( They were previously about 10-12 dB less ).  Overall, the volume was much better.  Just maybe a few dB less than a non-Dirac preset.  I somewhat expect this since it has to either cut or raise something.  I exported using the default Dirac curve and also exported another using the Harmon 4db for all speakers and the +6dB for the Center sub.  These new tests sound much better than the first test, but the whole thing still sounds really bad.  The room sounds like every inch of the walls and ceiling have been treated.  There is almost no ambient noise.  It sounds as if the dialogue has been extracted and is playing over top of a muted version of the rest of the track.  Everything sounds muffled but also discreet which makes no sense in how I'm trying to explain it.  It is really hard to explain what I'm experiencing.  The wife got in from a trip about an hour ago and walked into the room and said it sounded awful.

I've watched (3) Dirac videos in the past 2 hours. I've looked at all the target curves. Nothing seems to any different than what I'm seeing other people do.  Maybe I don't know what this stuff is actually supposed to sound like.  Maybe I'm so used to a room having some liveliness that I can't appreciate how it is supposed to sound.  I can tell you I don't like it even a little.  My other preset with only levels matched and crossovers set sounds pretty darn good.  It sounded better when I had some PEQ in, but I've kinda lost all that and need to rerun all those measurements in REW to have a PEQ based preset.  I guess what I'm saying is that between what I'm hearing with Dirac so far, I'd take no filters 100 times out of 100.  It is night and day different.

At any rate, thanks for asking and trying to help!  I'll keep trying to figure out what I'm doing wrong.

Well the latency from the mini DSP would make Dirac think you are in very large house, as in almost stadium size. So time alignment would most likely be an issue. As for the way it sounds, honestly, we have nothing to do with that. That is all on the Dirac side, all we do is take the coefficients and apply them. Now the volume is an offset Dirac comes up with based on the loudest speaker it measures, plus another 6db of headroom. So try to balance out the SPL of all your drivers so it doesn’t cause a wild level shift. 6db is normal, above that, it is offsetting for a particular speaker. The curve is something you will have to play with, but I am concerned about the results given the time lag associated with an additional DSP in the loop.

Lonnie

[ttocs]

Jul 26, 2020 18:06:21 GMT -5 Lonnie said:

Jul 26, 2020 15:52:39 GMT -5 megash0n said:

Thanks Lonnie .  
. . . . . . . . . . .
At any rate, thanks for asking and trying to help!  I'll keep trying to figure out what I'm doing wrong.

Well the latency from the mini DSP would make Dirac think you are in very large house, as in almost stadium size. So time alignment would most likely be an issue. As for the way it sounds, honestly, we have nothing to do with that. That is all on the Dirac side, all we do is take the coefficients and apply them. Now the volume is an offset Dirac comes up with based on the loudest speaker it measures, plus another 6db of headroom. So try to balance out the SPL of all your drivers so it doesn’t cause a wild level shift. 6db is normal, above that, it is offsetting for a particular speaker. The curve is something you will have to play with, but I am concerned about the results given the time lag associated with an additional DSP in the loop.

Lonnie

So it sounds like the process should be to remove all processing in the chain, run Dirac, then add back any post processing someone may want. Do I have this right?

[foggy1956]

Jul 26, 2020 18:09:49 GMT -5 megash0n said:

Jul 26, 2020 18:06:21 GMT -5 Lonnie said:

Well the latency from the mini DSP would make Dirac think you are in very large house, as in almost stadium size. So time alignment would most likely be an issue. As for the way it sounds, honestly, we have nothing to do with that. That is all on the Dirac side, all we do is take the coefficients and apply them. Now the volume is an offset Dirac comes up with based on the loudest speaker it measures, plus another 6db of headroom. So try to balance out the SPL of all your drivers so it doesn’t cause a wild level shift. 6db is normal, above that, it is offsetting for a particular speaker. The curve is something you will have to play with, but I am concerned about the results given the time lag associated with an additional DSP in the loop.

Lonnie

Thank you. I'll dig in more tomorrow.

I believe you have already time aligned your subs with your mains so time lag should not be an issue, correct? Apologies, I need to rethink this.😊

[Lonnie]

Jul 26, 2020 18:48:43 GMT -5 ttocs said:

Jul 26, 2020 18:06:21 GMT -5 Lonnie said:

Well the latency from the mini DSP would make Dirac think you are in very large house, as in almost stadium size. So time alignment would most likely be an issue. As for the way it sounds, honestly, we have nothing to do with that. That is all on the Dirac side, all we do is take the coefficients and apply them. Now the volume is an offset Dirac comes up with based on the loudest speaker it measures, plus another 6db of headroom. So try to balance out the SPL of all your drivers so it doesn’t cause a wild level shift. 6db is normal, above that, it is offsetting for a particular speaker. The curve is something you will have to play with, but I am concerned about the results given the time lag associated with an additional DSP in the loop.

Lonnie

So it sounds like the process should be to remove all processing in the chain, run Dirac, then add back any post processing someone may want. Do I have this right?

yes, you are dead on.

Lonnie

[megash0n]

Jul 26, 2020 18:52:33 GMT -5 foggy1956 said:

Jul 26, 2020 18:09:49 GMT -5 megash0n said:

Thank you. I'll dig in more tomorrow.

I believe you have already time aligned your subs with your mains so time lag should not be an issue, correct? Apologies, I need to rethink this.😊

They were on a non Dirac preset. I no longer have control of distance unless I'm missing something inside of Dirac.

[megash0n]

Jul 26, 2020 18:48:43 GMT -5 ttocs said:

Jul 26, 2020 18:06:21 GMT -5 Lonnie said:

Well the latency from the mini DSP would make Dirac think you are in very large house, as in almost stadium size. So time alignment would most likely be an issue. As for the way it sounds, honestly, we have nothing to do with that. That is all on the Dirac side, all we do is take the coefficients and apply them. Now the volume is an offset Dirac comes up with based on the loudest speaker it measures, plus another 6db of headroom. So try to balance out the SPL of all your drivers so it doesn’t cause a wild level shift. 6db is normal, above that, it is offsetting for a particular speaker. The curve is something you will have to play with, but I am concerned about the results given the time lag associated with an additional DSP in the loop.

Lonnie

So it sounds like the process should be to remove all processing in the chain, run Dirac, then add back any post processing someone may want. Do I have this right?

I'm not sure how you do that with multiple subs. Unless you are saying to just not measure the subs. I don't personally see how it is any different than using REW to align. Is there something about Dirac that is going to "see" more delay that a REW measurement?

[Lonnie]

Jul 26, 2020 18:52:33 GMT -5 foggy1956 said:

Jul 26, 2020 18:09:49 GMT -5 megash0n said:

Thank you. I'll dig in more tomorrow.

I believe you have already time aligned your subs with your mains so time lag should not be an issue, correct? Apologies, I need to rethink this.😊

Think about like this.

Dirac tells us to start to test pattern and the clock starts. We generate the test tones in the DSP, to the DAC, to the Amp, to the Speaker, to the Mic, clock stops. Now it knows how far away the speaker is.

His setup is like this. Dirac says go and the clock starts. We generate the test tones in the DSP, to the DAC, to the Mini DSP, Analog to Digital conversion, to the second DSP, through any Time Delays he set, to the second DAC, to the Amp, to the subs, to the mic, clock stops. Now it thinks the subs are way farther away from everything else. So it sets their time delay to 0 and pushes all the other channels that far away. The overall phase and time alignment would not be ideal.

Lonnie

[ttocs]

Just for fun I just wanted to show what happens if one speaker in a group is not quite right.

In swapping around my speakers I hadn't yet installed an in-wall speaker because there wasn't enough space within the outside wall to bury the speaker, so I needed to build an extension baffle for it. I wanted to see what would happen so I performed a measurement session. This is the result.

Left Rear Speaker sitting on counter aimed out like it would be when installed. No bass, rolled off treble. Timing is messed up! Notice that Dirac place the Curtains based on this "bad" speaker which affects the entire Group even though the other 3 were more capable.
    

This is the Right Rear Speaker. This is what the Left Rear should be close to in performance.
   

[foggy1956]

Jul 26, 2020 19:15:13 GMT -5 megash0n said:

Jul 26, 2020 18:48:43 GMT -5 ttocs said:

So it sounds like the process should be to remove all processing in the chain, run Dirac, then add back any post processing someone may want. Do I have this right?

I'm not sure how you do that with multiple subs. Unless you are saying to just not measure the subs. I don't personally see how it is any different than using REW to align. Is there something about Dirac that is going to "see" more delay that a REW measurement?

I think that while the minidsp will align the subs relative to each other, there is no way to align the subs relative to the mains if Dirac doesn't do it internally.

[Lonnie]

Jul 26, 2020 19:15:13 GMT -5 megash0n said:

Jul 26, 2020 18:48:43 GMT -5 ttocs said:

So it sounds like the process should be to remove all processing in the chain, run Dirac, then add back any post processing someone may want. Do I have this right?

I'm not sure how you do that with multiple subs. Unless you are saying to just not measure the subs. I don't personally see how it is any different than using REW to align. Is there something about Dirac that is going to "see" more delay that a REW measurement?

What is your system config? 5.2.4, 7.1.6, 9.1.6?

Lonnie

[ttocs]

Jul 26, 2020 19:12:44 GMT -5 megash0n said:

Jul 26, 2020 18:52:33 GMT -5 foggy1956 said:

I believe you have already time aligned your subs with your mains so time lag should not be an issue, correct? Apologies, I need to rethink this.😊

They were on a non Dirac preset. I no longer have control of distance unless I'm missing something inside of Dirac.

No you're not missing anything. Distance won't even show up once a Dirac Filter is active on the processor. I'm not aware of any Distance customization within Dirac.

This is before any Dirac filter is loaded onto the processor. Distance is shown as a setting, so is Name which is to customize the Name of the selected Preset.


This is what you'll see AFTER loading a Dirac Filter. Distance is gone, so is Name.

[megash0n]

Jul 26, 2020 19:21:25 GMT -5 Lonnie said:

Jul 26, 2020 19:15:13 GMT -5 megash0n said:

I'm not sure how you do that with multiple subs. Unless you are saying to just not measure the subs. I don't personally see how it is any different than using REW to align. Is there something about Dirac that is going to "see" more delay that a REW measurement?

What is your system config? 5.2.4, 7.1.6, 9.1.6?

Lonnie

5.1.4
Center sub feeds input 1 and that feeds two channels of a sub amp.

[megash0n]

Jul 26, 2020 19:18:05 GMT -5 Lonnie said:

Jul 26, 2020 18:52:33 GMT -5 foggy1956 said:

I believe you have already time aligned your subs with your mains so time lag should not be an issue, correct? Apologies, I need to rethink this.😊

Think about like this.

Dirac tells us to start to test pattern and the clock starts. We generate the test tones in the DSP, to the DAC, to the Amp, to the Speaker, to the Mic, clock stops. Now it knows how far away the speaker is.

His setup is like this. Dirac says go and the clock starts. We generate the test tones in the DSP, to the DAC, to the Mini DSP, Analog to Digital conversion, to the second DSP, through any Time Delays he set, to the second DAC, to the Amp, to the subs, to the mic, clock stops. Now it thinks the subs are way farther away from everything else. So it sets their time delay to 0 and pushes all the other channels that far away. The overall phase and time alignment would not be ideal.

Lonnie

Are you saying these things are exclusive to Dirac, or are you just saying, in general, that a mindsp will add latency to any measurement instrument? Meaning.. They will all have to deal with it the same. REW sees this latency and it's just part of the measurement.

[megash0n]

Jul 26, 2020 19:20:04 GMT -5 foggy1956 said:

Jul 26, 2020 19:15:13 GMT -5 megash0n said:

I'm not sure how you do that with multiple subs. Unless you are saying to just not measure the subs. I don't personally see how it is any different than using REW to align. Is there something about Dirac that is going to "see" more delay that a REW measurement?

I think that while the minidsp will align the subs relative to each other, there is no way to align the subs relative to the mains if Dirac doesn't do it internally.

how does it when it doesn't know anything about the crossover?

[foggy1956]

Jul 26, 2020 19:37:26 GMT -5 megash0n said:

Jul 26, 2020 19:20:04 GMT -5 foggy1956 said:

I think that while the minidsp will align the subs relative to each other, there is no way to align the subs relative to the mains if Dirac doesn't do it internally.

how does it when it doesn't know anything about the crossover?

Crossover is fairly irrelevant to time alignment.