Would a double room correction be a bad idea? If not - why don't we have it?

[Oddball]

I sometimes get ideas while taking a morning shower. IMO a good time to think as one is still not fully awake. Some ideas are decent, some good, and some are of course just plain stupid. Tell me what you think about this one.

Let's take one EQ system as example. Say Dirac with ART. You measure and then set parameters in the software and send the filters to AVR/AVP. Story ends, unless people choose to use REW or similar and measure what was actually corrected and how. Based on the measurements and set parameters you can also see what Dirac predicts as in room response.

Well, what if the story would not end here and Dirac could do a round two. It could measure corrected response and make further corrections to the extent necessary. My knowledge is practical and don't pretend to understand DSP world in depth and especially in theory, thus the question. If there would be ability to make additional corrections, would it, in theory at least, have the opportunity to improve its own results based on actual, rather than predicted results?

Some people do these corrections manually, but that is based on complicated combination of their experience and understanding of room acoustics that is not so common. As I understand some corrections would be to focused on what was asked from the EQ system but it could not deliver, and some to correcting over-corrections done by the software that might not be warranted. Automatic solution like this would probably require some additional code, but with EQ systems getting smarter, this might be an opportunity to further improve the results, or at least cater to the paranoia of the advanced users.

I paid pretty penny for Dirac ART (and consider it worth the expense), but would not mind parting with additional $100 for the double-up additional Dirac module. Would hope that such module would not be snake oil given that traditionally Dirac was dealing expensively but fairly.

 

[Keith_W]

You are talking about chaining two DSP's in series and wondering whether it would provide any additional benefit?

Whether this would work for you or not depends on how severe your problems are, and whether one DSP unit has sufficient biquads/PEQ's to deal with all your issues. For most people, ONE MiniDSP unit, with 10 biquads available for programming per channel, should be sufficient. There are people who use dozens of biquads, but I don't know enough about what they are doing to understand why they would want to do such a thing. If your post-Dirac correction appears satisfactory with REW, then there probably isn't much to be gained by adding more DSP processing.

That said, I can sense you have a certain itch that you want to scratch. Perhaps consider manual DSP correction? It's fun!

 

[ppataki]

I sometimes get ideas while taking a morning shower. IMO a good time to think as one is still not fully awake. Some ideas are decent, some good, and some are of course just plain stupid. Tell me what you think about this one.

Let's take one EQ system as example. Say Dirac with ART. You measure and then set parameters in the software and send the filters to AVR/AVP. Story ends, unless people choose to use REW or similar and measure what was actually corrected and how. Based on the measurements and set parameters you can also see what Dirac predicts as in room response.

Well, what if the story would not end here and Dirac could do a round two. It could measure corrected response and make further corrections to the extent necessary. My knowledge is practical and don't pretend to understand DSP world in depth and especially in theory, thus the question. If there would be ability to make additional corrections, would it, in theory at least, have the opportunity to improve its own results based on actual, rather than predicted results?

Some people do these corrections manually, but that is based on complicated combination of their experience and understanding of room acoustics that is not so common. As I understand some corrections would be to focused on what was asked from the EQ system but it could not deliver, and some to correcting over-corrections done by the software that might not be warranted. Automatic solution like this would probably require some additional code, but with EQ systems getting smarter, this might be an opportunity to further improve the results, or at least cater to the paranoia of the advanced users.

I paid pretty penny for Dirac ART (and consider it worth the expense), but would not mind parting with additional $100 for the double-up additional Dirac module. Would hope that such module would not be snake oil given that traditionally Dirac was dealing expensively but fairly.

I have always done room correction in 2-passes, well, actually 4-passes in total

1. First I apply pre-EQ in the MLP in order to make all 4-ways more or less flat (ideally 2-4 filters shall be enough on top of the crossover filters of course)
2. Then I apply Dirac Live
3. Then I re-measure what Dirac did in the MLP and I apply post-EQ to flatten the response (Dirac does not make the response totally flat, not even if I set its target curve to flat)
4. Finally I apply a 'target curve' to taste (basically a low shelf and a high shelf)

I could apply the target curve in Dirac already to reduce the number of steps but I don't like the sound of it, I prefer to create my own target curve
I have been using this for 10 years and it works really well (for me)

 

[Oddball]

You are talking about chaining two DSP's in series and wondering whether it would provide any additional benefit?

Whether this would work for you or not depends on how severe your problems are, and whether one DSP unit has sufficient biquads/PEQ's to deal with all your issues. For most people, ONE MiniDSP unit, with 10 biquads available for programming per channel, should be sufficient. There are people who use dozens of biquads, but I don't know enough about what they are doing to understand why they would want to do such a thing. If your post-Dirac correction appears satisfactory with REW, then there probably isn't much to be gained by adding more DSP processing.

That said, I can sense you have a certain itch that you want to scratch. Perhaps consider manual DSP correction? It's fun!

Thanks Keith. Not talking about daisy chain. Talking about hypothetical case where Dirac ART, or other system, could take another set of measurements once correction is done and filters loaded, and correct the filters again based on the actual in-room response of original filters. Something that people like @ppataki are doing, only automated.

As getting older my enthusiasm seems to be waning across the board. I have done my share of manual and semi-manual EQ when thought it was fun. Would not go back there - at least at this point. Aware that i should not ever say never as that is a mighty long time

 

[Oddball]

I have always done room correction in 2-passes, well, actually 4-passes in total

1. First I apply pre-EQ in the MLP in order to make all 4-ways more or less flat (ideally 2-4 filters shall be enough on top of the crossover filters of course)
2. Then I apply Dirac Live
3. Then I re-measure what Dirac did in the MLP and I apply post-EQ to flatten the response (Dirac does not make the response totally flat, not even if I set its target curve to flat)
4. Finally I apply a 'target curve' to taste (basically a low shelf and a high shelf)

I could apply the target curve in Dirac already to reduce the number of steps but I don't like the sound of it, I prefer to create my own target curve
I have been using this for 10 years and it works really well (for me)

Thanks - you of course have immensely more knowledge than most members. I was mentioning steps 2-4, as ideally step 1 would be somehow integrated in the other 3 steps. Your system and approach is one of the reason for my question that would cater to less adventurous members that would not mind double calibration run - and of course third to verify with REW.

 

[ppataki]

ideally step 1 would be somehow integrated in the other 3 steps

Yes, I totally see your point - the reason for step 1 (in my case, at least) is that I use closed cabinet speakers only (DIY, see in my signature) and those need a low-shelf compensation in the low end. Dirac not always can do that so by providing it with a pre-low-shelf filter it can successfully linearize the response afterwards in step 2
For people using 'normal' speakers I guess step 1 can be safely omitted.

 

[Mort]

This is what Genelec recommends if using third party DSP with glm. First to flatten, then the curve.

In practice I didn't find much improvement, often was worse.

 

[Ogoript]

I tried this on my Wiim Ultra. I EQ'd my Kef LS50 Wireless II based on the recommended spinorama PEQ. Then I ran Roomfit to correct impact of the room. It seemed to work pretty well. I haven't taken any measurements to verify it, but t he EQ based on spinorama made quite a difference in general tonality. The roomfit in Wiim correct the bass bloom from my subs.

 

[sam_adams]

I sometimes get ideas while taking a morning shower.

 

[90sNintendoKid]

I’ve been thinking of trying the same thing.

I’ve got a Minidsp Flex HTX and I’m thinking of correcting the speakers to my room’s target curve and then letting Dirac or A1 help integrate my speakers and subwoofers and do a second pass.

I have used the Magic Beans 2 pass Dirac method with ART but I’ve had problems with ART no hitting my target curve on the subwoofers, which I admit could be a user error on my part.

I’ve still got experimenting to do.