AVS Forum banner

4261 - 4280 of 4511 Posts

·
Registered
Joined
·
3,083 Posts
I consider sub PEQ and a "house curve" as different things. Applying PEQ prior to the calibration is reasonable. Applying a house curve is best done with a custom target after the calibration.
I do 2 things with the miniDSP:
1. I take a combined sub measure in REW and use the EQ tab to add a room curve and then export and load it into the miniDSP output PEQ
2. I remeasure and then go into the input PEQ of the miniDSP and adjust peaks/dips more manually, measuring between changes

Are both of these what you consider to be PEQ? Which or both should I still perform before running DL?
 

·
Registered
Joined
·
487 Posts
Got these surprising results with REW after running DLBC on HTP-1 with 7.4 setup in large room. Dirac set crossovers at 70 hz. The first image is Dirac with a Harman +10 curve. Got the curve from beginning of this thread. I don't see the curve doing what you would expect. In fact, I don't see it doing anything at all.
The second image is after raising sub trims (in subs) by 10 dbs. Then, raised dbs in Dirac either 4 or 6 at 20 hz. As you can see, the 4 hz. raise includes a null at 60hz of almost 9 dbs. The 6 hz raise does not.
Very happy with results of Dirac and the +6 at 20hz settings. Just mystified by REW results.
I know just enough to confuse myself, but not enough to even start to understand these results.
Dirac Harman +10.jpg
+4 vs +6 dbs at 20hz.jpg
 

·
** Man of Leisure **
Joined
·
24,611 Posts
I do 2 things with the miniDSP:
1. I take a combined sub measure in REW and use the EQ tab to add a room curve and then export and load it into the miniDSP output PEQ
2. I remeasure and then go into the input PEQ of the miniDSP and adjust peaks/dips more manually, measuring between changes

Are both of these what you consider to be PEQ? Which or both should I still perform before running DL?
I suspect that when you say "add a room curve", you mean that you check the "add room curve" on the REW EQ page. In my measurements, I get the same predicted output curve whether I check "add room curve" or leave this option unchecked. So in either case, I would consider this PEQ. Whether you get better final results by adding PEQ before you run the Dirac calibration or not would depend on a number of factors. The only way you could tell would be to try running Dirac with and without PEQ and measure the results.

Here is what I am seeing:

REW PEQ tool showing effectiveness of PEQ in the Predicted curve:

3119969


3119970


3119973
 

Attachments

·
Registered
Joined
·
3,083 Posts
I suspect that when you say "add a room curve", you mean that you check the "add room curve" on the REW EQ page. In my measurements, I get the same predicted output curve whether I check "add room curve" or leave this option unchecked. So in either case, I would consider this PEQ. Whether you get better final results by adding PEQ before you run the Dirac calibration or not would depend on a number of factors. The only way you could tell would be to try running Dirac with and without PEQ and measure the results.

Here is what I am seeing:

REW PEQ tool showing effectiveness of PEQ in the Predicted curve:

View attachment 3119969

View attachment 3119970

View attachment 3119973
Very good. Thanks for this. I will experiment in my next measurement session.
 

·
Premium Member
Joined
·
699 Posts
Got these surprising results with REW after running DLBC on HTP-1 with 7.4 setup in large room. Dirac set crossovers at 70 hz. The first image is Dirac with a Harman +10 curve. Got the curve from beginning of this thread. I don't see the curve doing what you would expect. In fact, I don't see it doing anything at all.
The second image is after raising sub trims (in subs) by 10 dbs. Then, raised dbs in Dirac either 4 or 6 at 20 hz. As you can see, the 4 hz. raise includes a null at 60hz of almost 9 dbs. The 6 hz raise does not.
Very happy with results of Dirac and the +6 at 20hz settings. Just mystified by REW results.
I know just enough to confuse myself, but not enough to even start to understand these results.
View attachment 3119942 View attachment 3119947
FWIW, re-running DLBC will give different results with each calculation, even if you use the same measurements, so it might be worth rerunning the calculation if you end up with what seems to be bizarre results (like an uncharacteristic null). Example: if you make a set of measurements and then save that raw data and then add a target curve and set the crossover and curtain, Dirac will do its calculation and produce the filter to save to your processor/AVR. Measure the results in REW. Then, go back to Dirac and open up the original saved raw measurement data and rerun the drill (that is, set the identical crossover and curtain), then export the new filter and rerun REW. You’ll find that the results will not be identical. Sometimes you might end up with a bizarre null. When that happens, try doing the calculation again. There seems to be a bit of chance with the calculations.
 

·
** Man of Leisure **
Joined
·
24,611 Posts
Thanks to all of you and your continued troubleshooting of this platform. So are we still operating under the "Don't use DLBC until further notice"?
We are still waiting for a fix on the problem that was reported earlier.
 

·
** Man of Leisure **
Joined
·
24,611 Posts
I'm still using it. Why? I like the way it sounds,to me.
There is nothing wrong with continuing to use DLBC as long as you know the nature of the issue, and when it manifests itself.
 

·
Registered
Joined
·
159 Posts
After cancelling my DLBC license and trying to replace it with MSO+MiniDSP+regular DiralLive, I have observed two things:

MSO does a wonderful job optimizing the bass response throughout my two rows theater, something DiracLive with BassControl never managed to achieve, both subjectively and as measured with REW. No matter what mic placement I tried in DiracLive, I never got my second row to sound good. Now, with MSO, I cannot believe the bass as it sounds almost the same in every seat!

But without DLBC, I never was able to achieve the same level of sub+mains integration. Using MSO, DiracLive regular and a MiniDSP, I tried hard to make my mains blend well with my subs. I got it good, but not great. It is my opinion that DLBC does wonders there. It is easy, fully automated and yields great results.

So yesterday, after reading Flax's post, I decided to purchase the DLBC license again. This time, I purchased the single sub license instead of the multisub. I recalibrated everything, keeping my subs connected to the MiniDSP and calibrated as "one" with MSO. My system never sounded so great! Subwoofer output has never been so equal in every seat and stereo music never sounded go great!

Of course, Now that I know the bug is there, I cannot help but notice the obvious bass degradation as soon as I switch from stereo to Auro upmix. But now, I am confident that Dirac will fix it.

As a bonus of running my sub as "one", I freed two channels on my HTP-1 and can now use them for my front wides instead of my 2nd and 3rd sub.
Great post on replacing multi sub DLBC with MSO/MiniDSP/regular Dirac or single sub DLBC.

I’m strongly considering doing the same and bumping up to using wides for 9.3.6.

The longer I’ve waited for a fix, the more this option appealed to me. Then I read up and saw some talks about speaker layouts by Anthony Grimani, the Trinnov guys, and others and see the value in wides. And there’s lots of new Atmos content coming out, and the DTS:X Pro upmixer should be more widely supported soon.

Originally, I thought DLBC wold save me a ton of time. But my room seems to be a tougher edge case. After dozens and dozens of Dirac DLBC optimizations with multiple placement changes, I could not get good seat to seat consistency over a single row.

I posted about my problem earlier in this thread, then opened a Dirac help ticket. They analyzed my Dirac project file and REW measurements and said that it looks like the algorithm is not working great with my room and that a future update to the algorithm might fix it all up. They kindly offered a refund at that time but I decided to wait for software improvements.

I also asked if “Pre-EQing” my subs before running Dirac would help and they said yes it would help Dirac do its job better because it has eq limits. I had planned on doing this in my HTP-1 but decided to wait on some speaker upgrades and possible Dirac updates. (Most people don’t need to pre-EQ by the looks of it but can come in handy in some cases.)

And now, this bug—at this point, I might just bite the bullet and do the EQ + MSO work, using a hopefully soon to be fixed DLBC to splice the “single sub” with all of the surrounds.

The downsides are that I need to buy another device for DSP, plus a windows machine to run MSO. And I need to do all the work I was hoping DLBC would save me, but after dozens of Dirac plus REW measurement runs I’ve already burned through that budget :)

The upside is I get 2 more channels for wides and I can have unlimited subs :cool:
 

·
** Man of Leisure **
Joined
·
24,611 Posts
I think we all face the same decision--use DLBC, or optimize using a single sub connection and MSO. If we had a working DLBC solution, we could make a decision based on facts and data. Unfortunately, that is not the case (for now).
 

·
Registered
Joined
·
159 Posts
I think we all face the same decision--use DLBC, or optimize using a single sub connection and MSO. If we had a working DLBC solution, we could make a decision based on facts and data. Unfortunately, that is not the case (for now).
Indeed.

BTW, I just your signature and that you authored the MiniDSP how to optimize multi subs & the REW guide, referenced in the MSO docs. You and @markus767 are in the vanguard!

I'm sure Dirac will have a DLBC fix. I guess we just need to remain patient. I'm ok with waiting, but the pause did give me time to think about upgrades. I would still like to use DLBC, even if it's just for a "single" MSO optimized sub array.
 

·
Registered
Joined
·
487 Posts
FWIW, re-running DLBC will give different results with each calculation, even if you use the same measurements, so it might be worth rerunning the calculation if you end up with what seems to be bizarre results (like an uncharacteristic null). Example: if you make a set of measurements and then save that raw data and then add a target curve and set the crossover and curtain, Dirac will do its calculation and produce the filter to save to your processor/AVR. Measure the results in REW. Then, go back to Dirac and open up the original saved raw measurement data and rerun the drill (that is, set the identical crossover and curtain), then export the new filter and rerun REW. You’ll find that the results will not be identical. Sometimes you might end up with a bizarre null. When that happens, try doing the calculation again. There seems to be a bit of chance with the calculations.
Thank you for your response. That's as good an explanation as I can think of. Still curious whether anyone else has tested Harman curve with REW and been disappointed. Our ears confirm REW results inasmuch as raising sub trim in subs with additional bump at 20 hz give much better results than the Harman +10. At any rate, we are very happy with the results. I stopped running REW after getting DLBC because results are so good. Just was surprised when I did run out of curiosity.
When I had ARC Genesis, I always re-adjusted phase and level matching for the 4 subs after raising their volume post calibration. A very knowledgeable, but not audiophile, friend told me that whenever you change one setting, even volume, you are effecting everything. With DLBC, the integration of subs is beyond anything I was able to achieve using ARC Genesis and REW. However, because of the precision of the sub measurements, I cannot see how I could adjust their delays or relative volumes without destroying the results. Obviously by changing sub trims in subs after calibration I am effecting something else, resulting in the null with the +4 at 20hz setting, but not the +6 alternative.
Thanks again for your response.
 

·
Registered
Joined
·
683 Posts
After cancelling my DLBC license and trying to replace it with MSO+MiniDSP+regular DiralLive, I have observed two things:

MSO does a wonderful job optimizing the bass response throughout my two rows theater, something DiracLive with BassControl never managed to achieve, both subjectively and as measured with REW. No matter what mic placement I tried in DiracLive, I never got my second row to sound good. Now, with MSO, I cannot believe the bass as it sounds almost the same in every seat!

But without DLBC, I never was able to achieve the same level of sub+mains integration. Using MSO, DiracLive regular and a MiniDSP, I tried hard to make my mains blend well with my subs. I got it good, but not great. It is my opinion that DLBC does wonders there. It is easy, fully automated and yields great results.

So yesterday, after reading Flax's post, I decided to purchase the DLBC license again. This time, I purchased the single sub license instead of the multisub. I recalibrated everything, keeping my subs connected to the MiniDSP and calibrated as "one" with MSO. My system never sounded so great! Subwoofer output has never been so equal in every seat and stereo music never sounded go great!

Of course, Now that I know the bug is there, I cannot help but notice the obvious bass degradation as soon as I switch from stereo to Auro upmix. But now, I am confident that Dirac will fix it.

As a bonus of running my sub as "one", I freed two channels on my HTP-1 and can now use them for my front wides instead of my 2nd and 3rd sub.
That’s interesting. I’ve been weighing the decision of adding wides vs. using DLBC as someone just posted above. I’m curious on what people think about using this method of MSO with miniDSP and DLBC single sub. I’m wondering If it makes sense that MSO and DLBC would work well together.

Also, MSO is a windows only program if I remember correctly. I have no idea how it works or how it’s implemented but I would have to use it on my iMac using boot camp. Curious if there’s an issue with that.
 

·
Registered
Joined
·
5 Posts
Just to expand a bit on the topic, below are two diagrams illustrating the use of all-pass filters in a bass managed system. This isn't specific to DL but applies to any room/speaker correction attempt when using all-pass filters for crossover splice optimization (in MSO for example).

View attachment 3118261

View attachment 3118262
Seems to me that we have to beyond even the second picture to solve the problem.

Any deviation from mono (within the DLBC frequency band, including both subs and mains) interferes with the solution to some degree, since the differing channel contributes too much or too little to the global phase solution. Turning on only a couple channels as you did (e.g. L+C or L+RS) demonstrates that problem to the maximum extent.

How about if DLBC works like this instead. Have a user-selectable cutoff frequency, the "direction matters cutoff". Below that frequency, use the DLBC algorithm but make sure all input content is pre-combined into one mono signal (upstream of the HPF/LPF in your pic) that gets fed to all channels. Above that frequency, use just the regular dirac-live-style independent correction.

I would envision the range of this cutoff selection to be perhaps 60-500 Hz, depending on room size/modes.
 

·
** Man of Leisure **
Joined
·
24,611 Posts
Seems to me that we have to beyond even the second picture to solve the problem.

Any deviation from mono (within the DLBC frequency band, including both subs and mains) interferes with the solution to some degree, since the differing channel contributes too much or too little to the global phase solution. Turning on only a couple channels as you did (e.g. L+C or L+RS) demonstrates that problem to the maximum extent.

How about if DLBC works like this instead. Have a user-selectable cutoff frequency, the "direction matters cutoff". Below that frequency, use the DLBC algorithm but make sure all input content is pre-combined into one mono signal (upstream of the HPF/LPF in your pic) that gets fed to all channels. Above that frequency, use just the regular dirac-live-style independent correction.

I would envision the range of this cutoff selection to be perhaps 60-500 Hz, depending on room size/modes.
Options like you are suggesting add a layer of complexity to the DLBC solution, most likely beyond the average user's level of expertise. The compelling reason to use DLBC is that it should provide excellent results as automatically as possible.
 

·
Premium Member
Joined
·
699 Posts
Thank you for your response. That's as good an explanation as I can think of. Still curious whether anyone else has tested Harman curve with REW and been disappointed. Our ears confirm REW results inasmuch as raising sub trim in subs with additional bump at 20 hz give much better results than the Harman +10. At any rate, we are very happy with the results. I stopped running REW after getting DLBC because results are so good. Just was surprised when I did run out of curiosity.
When I had ARC Genesis, I always re-adjusted phase and level matching for the 4 subs after raising their volume post calibration. A very knowledgeable, but not audiophile, friend told me that whenever you change one setting, even volume, you are effecting everything. With DLBC, the integration of subs is beyond anything I was able to achieve using ARC Genesis and REW. However, because of the precision of the sub measurements, I cannot see how I could adjust their delays or relative volumes without destroying the results. Obviously by changing sub trims in subs after calibration I am effecting something else, resulting in the null with the +4 at 20hz setting, but not the +6 alternative.
Thanks again for your response.
That is pretty interesting. One thing you might try is running the RTA function in REW and fiddling with the sub trim to see if you can get rid of that null, if you want to try and avoid remeasuring and all that jazz. You might find some luck there with even a minuscule adjustment. Any kind of adjustment can result in some pretty radical changes (and oftentimes, hardly any changes at all), and running RTA can reveal that. Bass frequencies are weird but freaking cool. One caveat: the adjustments will affect areas (that is, if you’re adjusting with the mic at the listening position, you won’t know how that is affecting the seat to your left, behind, etc.).
 

·
Registered
Joined
·
159 Posts
That’s interesting. I’ve been weighing the decision of adding wides vs. using DLBC as someone just posted above. I’m curious on what people think about using this method of MSO with miniDSP and DLBC single sub. I’m wondering If it makes sense that MSO and DLBC would work well together.

Also, MSO is a windows only program if I remember correctly. I have no idea how it works or how it’s implemented but I would have to use it on my iMac using boot camp. Curious if there’s an issue with that.
Below is an exerpt from the MSO intro page

I don’t see a reason it wouldn’t work well with DLBC to “optimize the splice” and it looks like it works well as @stef2 noted. There probably not a ton of people that have used this combo because DLBC is supposed to automate the sub array optimization process as well. So we only have early adopters waiting for a bug fix using this combo, or people that wanted to optimize a sub array larger than their DLBC capable outputs.

MSO should work fine in Bootcamp. It just crunches numbers after importing REW measurements. You can still use your existing Mac REW rig to do the measurements if you want to. You can test MSO out with the provided tutorial measurements.

3120335
 

·
Registered
Joined
·
14,328 Posts
Discussion Starter · #4,279 · (Edited)
Seems to me that we have to beyond even the second picture to solve the problem.

Any deviation from mono (within the DLBC frequency band, including both subs and mains) interferes with the solution to some degree, since the differing channel contributes too much or too little to the global phase solution. Turning on only a couple channels as you did (e.g. L+C or L+RS) demonstrates that problem to the maximum extent.

How about if DLBC works like this instead. Have a user-selectable cutoff frequency, the "direction matters cutoff". Below that frequency, use the DLBC algorithm but make sure all input content is pre-combined into one mono signal (upstream of the HPF/LPF in your pic) that gets fed to all channels. Above that frequency, use just the regular dirac-live-style independent correction.

I would envision the range of this cutoff selection to be perhaps 60-500 Hz, depending on room size/modes.
What you describe is Dirac Unison — Dirac ;) The problem I see with that approach is that satellite speakers can't be part of an optimization strategy that is based on downmixing multichannel content to a single monophonic signal. This works only below about 80Hz where humans can't localize sound sources. Output of most satellite speakers is also highly limited in that range. I don't think it's a good idea to send LFE and all other bass content to them.

So fixing the current DLBC implementation would be good enough for me right now.
 

·
Registered
Joined
·
1,516 Posts
Thanks to all of you and your continued troubleshooting of this platform. So are we still operating under the "Don't use DLBC until further notice"?
From what I understand (and hear) DLBC is working just fine for 2.1 listening. The problem is with multichannel formats.
I enable DLBC 100% of the time when listening to stereo music, no bug there and sounds wonderful to my ears!
 
4261 - 4280 of 4511 Posts
Top