TOPIC:

Ultrasonic wrote:

hannesb wrote:

A correct centered soundstage always sounds better, more natural  

Errr, I never suggested having a correctly centered soundstage wasn't preferable. What was the answer to my last question? I'm trying to understand what may be going on... 

For information, it's possible to use REW to measure arrival time differences between the two speakers by making use of a consistent acoustic timing reference. So if you wanted to see what Dirac or anything else is doing in this regard you can. It might actually be informative to make measurements in REW like this without DL on but with your 0.44 ms delay applied to give your central image. Make measurements for each speaker separately, and if you share the mdat file here I'll have a look. I'm thinking seeing measurements of what sounds central to you may be informative... @asx77 has made a good point re. level balance rather than time delay being important for centering by the way.

I followed this approach: www.minidsp.com/applications/acoustic-me...ts/292-rew-timing-ht, but never gets a time delay of 0 in case of the same speaker for reference and output. I first did 2 measurements without any "corrections", the last 2 measurement imply a delay of 0.44ms for the right channel. No dirac live applied

Playing with the balance doesn't really work because I need to lower the gain on the right channel to much it is affecting the soundstage negatively


I can't upload the mdat file as it is exceeding the maximum allowed file size. I added 2 sreenshots related to my first measurements

Regarding my first measurements I get the following delays:
Left -0.1045 ms
Right - 0.2402

2nd measurements:
Left -0.1244 ms
Right: 0.4776

The easiest way to share mdat files is to host them somewhere else, like Google Drive.

Ultrasonic wrote:

OK, so if you do a full Dirac measurement set, save filters, and then manually add your apparently preferred 0.44 ms delay to the required output channel, how does it sound?

^Please can you answer this question?

IUltrasonic wrote:

Ultrasonic wrote:

OK, so if you do a full Dirac measurement set, save filters, and then manually add your apparently preferred 0.44 ms delay to the required output channel, how does it sound?

^Please can you answer this question?

I think I found the solution regarding the center issue I'll need to listen more to make sure. Based on  the REW measurements, I adjusted time delay for both channels to bring it to 0 which is much lower than the 0.44ms on the right channel only. Then lowering the gain by around 2db on the right channel seems to do the trick. Soundstage reproduction makes more sense now. 

If I have time, I'll do some other testing with dirac live during the weekend. 

Thank you very much for all your help so far!

I'm pleased it sound like you're making progress. For what it's worth the pictures you shared made me think 0.44 ms was too much too but I'm always more confident when I can actually see the data for myself, and particularly look at the impulse response graphs.

For context I'll mention that I unfortunately have an asymmetric setup myself. If you haven't experimented already I'll just add that some speaker toe-in may help with central imaging if you haven't tried this already.

I would like to share my experiences because I have the same problem as hannesb:

I have an asymmetrical room (wall on the right, open on the left) and listen in the near field (distance approx. 1.1 m). I have a very tight listening position. The speakers are very, very exactly the same distance from the listening position (accurate to within a few millimeters).

Dirac adds a delay of 0.1 ms for the right channel, which means that the central image shifts a little to the right, but it is very noticeable. (By the way, I find it amazing what a strong effect a delay of only 0.1 ms = 34 mm has...) I "corrected" the delay added by Dirac by compensating it via the delay in the "Output" tab, i.e. I added a further delay of 0.1 ms to the left channel. The central image is now nicely centered. However, I'm not sure if my solution really represents the optimum, since it's somehow just a workaround...