So if it's virtualized speakers, like a phantom center, I assume the listening position must be very precise, no?
No. Somewhat yes, but mostly no.
Think about a phantom center and how far you can sit off center and still have the center image be reasonable. Now imagine a LCR where everything is perfect. How much more can you move while still having the anchoring effect?
Now imagine that your center speaker is a bit too far to the right. You need to pull the sound toward the left and imagine I mix a bit of the center into the left.
If you sit at the far left, the center is too far right from ideal but the faint additional signal from the left will pull the image closer to left. now imagine you are at true center. The center speaker is louder and dominant but the left speaker will pull you a bit to the left.
Now think about this:
Are your left and right speakers slightly closer to you than your center? Or is your LCR all in the same row? If your speakers are in a row, theoretically playing a bit of sound back right should barely move your main speaker position but you have to be careful so as not to bleed too much audio.
Now imagine that with phase knowledge, group delay , and EQ what you could do to minimize counterproductive bleed…
So yes, it’s not infinite Star Trek holodeck technology but the more speakers you have the better.
NEXT scenario, and this is used by Sonos.
Imagine I have a good set of speakers which are flat to 80Hz and roll off to - 3 dB at 55Hz. I have a left channel signal asking for 55Hz at SPLs greater than my left channel can achieve. What happens if I play the content on the right speaker? Since our ability to localize SPLs below 80Hz is poor and definitely at 55Hz, it’s even worse, I can get the target response.
Sonos sound bars will play non directional bass on *all* LCR+height drivers to get a lot of bass from a small series of woofers.
Now imagine you have a sub. And your sub fills in the 55 Hz just fine. But it is actually at 20 Hz, that your sub is -3 dB from target. What if you mixed in that content around 20 Hz to every other speaker in your setup?
Now you get the idea.
You need to know the phase and frequency response of every speaker. You need to understand the 3D position of every speaker, not just distance. You need to be able to know how best to throw information around to get improved sound field rather than loss of precision. Maybe the corrections only works if you correct 10% rather 100% of the mismatch, etc.
The Sony HT-A9 works great. This was their proof of concept.
The Trinnov in the Sherwood R972 works great.
The STR-AZ is a wild card but it seems like it really does work for the rooms and speakers some of the early adopters have.