Originally Posted by anothermib
I started to experiment with it a little, and it appears that an additional 3ms delay indeed lets the center blend in much better. I only tested it with a few tracks so far and need to see over time if it works with the bulk of my material, but so far it actually sounds pretty pleasant.
did you come up with any explanation why the front stage benefits from that additional delay? Is it dependent on the dimensions of the setup?
Not to any degree of certainty. My theory is that when the three L/C/R speakers are carrying identical audio (this is what happens in the center spread mode), and all three are time aligned, their sum is "fragile" in that if one moves slightly to the left of dead center, there are now 3 arrivals at 3 times. The sum is exhibiting a change in the constructive and destructive (combing) interference that we can easily detect. Considering it is a lot easier to acoustically match the L/R speakers than to match C to L/R, the chance of perfect summation is nil. By adding the C delay, the sensitivity to summation imperfections is moved to a lower range of frequencies, where the worst effects through the midrange are mitigated (delay is a form of decorrelation).
In my case e.g. the 3ms roughly aligns with the distance between center and mains, but that may be a coincidence.
The added delay in center is intended to be relative to coincident arrival time from L/C/R speakers. So whatever delays may be present in the normal calibration are purely coinky-dinky (hat tip to
Unfortunately my AVR doesn’t support multiple speaker configurations (nor PLIIx for that matter). Delays and speaker designations remain fixed, when changing between modes. So i need to find an optimal compromise (if that isn’t an oxymoron). I hope the additional delay doesn’t impact Atmos too much. I didn’t notice any degradation e.g. of dialogue intelligibility, but I will try out some of the Atmos demos.
Try to find something where a sound pans across L/C/R. Otherwise the difference will be mostly immaterial.
I am wondering if that kind of incremental delay may be beneficial for other, similar parts of the setup e.g. the side surrounds. Isn’t that a very comparable situation where they have to blend in between the front and rears?
In "sparse" speaker setups we use at home, the chances of 3 or more speakers carrying identical content is pretty slim. Usually we will see pair-wise panning. Of course now with immersive systems we can have objects positioned between 3 or 4 speakers, but in those cases the sound is usually short lived, and may be moving, so the chances of detecting timbral anomalies is quite low.
In larger home theaters with multiple side surrounds, it is not uncommon to provide some form of decorrelation (added delays or all-pass filters) to stagger the arrivals to prevent comb filtering, but that only arises because additional pairs of surrounds are fed the same signals.
I would, however, add that in my 7.1.4 setup I did offset one pair of the ceiling speakers by a few ms because I found that when Dolby Surround upmixes applause type sounds (anything sustained like that or rain), the front pair and rear pair are fed the same signals, so the combing becomes audible, especially if one moves during that time.
This has not been a problem in Atmos content, however, as the 4 height and even the 4 surround channels are properly created to avoid that.