TOPIC:

Does Dirac algorithm understand the mic positions?

I understand that first measurement from MLP Diract uses as an exact location: Delays and levels will
be calibrated so that this place is in the center.

But what about all other positions in various patterns? Does it understand that this signal comes now from
top left corner or from near left side, or are they just another separate values that be summed up into average?

I think I’ve read somewhere from Dirac that the software only cares about the centre position. You can do the others wherever you like and in whatever order. The image is just to give the user some idea. 

I think I’ve read somewhere from Dirac that the software only cares about the centre position. You can do the others wherever you like and in whatever order. The image is just to give the user some idea. 

Thanks for comment ASX77, that was quite same I wondered.

Let’s put it this way:

I want to have perfect sound in MLP, I sit 99%:ly in same place in a sofa, in MLP.

I usually take only 3 measurements: MLP first, then 30cm left from MLP, then 30cm right from MLP,
and that’s it. Dirac will now average these three and set levels/delays based on MLP. I expect to
get clearest sound and best correction now in my MLP.

Now some would say: You don’t give enough data for Dirac to analyze your room! OK… If I measure also
in front and in back from MLP, and from faaaar left and right, what would I give to Dirac? More wrong values
from locations where I don’t listen, and now it sums up these ”wrong” values into same average, and it
actallu ruins the best imaging in MLP..? 

Example: My sofa is 1,7m out from back wall. That equals to 1/4 wavelenght for approx 50Hz area, so it’s obvious
that I have bad cancellation at around 45-55Hz. (Thus I have fixed this with 80Hz x-over for mains and sub in other location).
Behind MLP this cancellation is a bit higher due to the wavelenght, so if I measure a bit back from MLP, I just give wrong
info to Dirac, and it might try to make unnecessary boost on 60Hz level, which we don’t need in MLP… Am I right or totally lost..? 😊

I haven't seen any clear documentation on exactly what the multiple measurement points do.  One possibility is that the multiple points give a better idea about what's possible to correct, and what isn't.  For example if there's a deep null across multiple points, not much will be gained by boosting that frequency.

I've always followed the Dirac recommendation on measuring 9 points, just because Dirac presumably know why they wrote the instructions that way, and if it was better to do 3, they'd say so.

I would encourage you to experiment with using more measurement points and then decide what you enjoy most on the basis of listening.

The exact locations of the Dirac measurements are not critical but part of the point of the range of the positions is to help the algorithm better handle the fact that some sound arrives directly from the speakers and some reflected off room boundaries. 

asx77 wrote:

I think I’ve read somewhere from Dirac that the software only cares about the centre position. You can do the others wherever you like and in whatever order. The image is just to give the user some idea. 

Very much correct assumption/statement from ASX77 here. Only the center position matters. Others are just a "GUI" representation for dirac live. 
3 would be the bare minimum I'd say.. Try to get more and you'll get better results in most case (i.e. trying to remove some outlier). Sometimes even out of these 3, one could have a bad response due to a room mode. Getting more measurements will average out and remove the outlier.. 

That's just our 2cts based on all measurements we've made but every does have his own technique/methods so as long as you're happy with the results, all good!  

devteam

Thanks for answer, this was just what I expected 😊👍🏻 So measuring ”far left” doesn’t tell to Dirac how the room sounds far left, it just adds one more average value.

Now I know how to work. I know my room quite well, as I have measured it all around with UMIK1 and REW.

I want to get best Imaging exactly in my MLP, so I prefer having 3-5 measurements tightly around MLP.