Few ways to go about this, his is one, your's is another, I suggest headphones and put an end to it all!!...........BUT,
The amp's output (speakers) are isolated so as long as the input power to the amp difference is 12V it wouldn't matter where the negative is referenced as long as the positive is only 12V above that neg ref point, it's fine.
NOW, what about run time/power?
If for example he goes as his suggested route implies with say 2-12V 50 a/h batts for an example, in series and hooked to the DC-DC converter (24--->12)
Then P(in) * eff = P(out)....@ *85% eff and where P = V*I
24*50*.85 = 12*x
Solve for x
x = 85 a/h @ 12V
Doing what he has proposed allowed him to run roughly and "continuous" about an hour if his amp, on avg, uses **85 amps with no battery switch out/over before the batts are 10.5V ea. (0[zero] SOC)
HOWEVER....and still using 0(zero) SOC as the baseline for this example.
He could just hook up to each battery one at a time sorta like you suggested, he'd gain 15% more runtime from not having the loss of the DC-DC AND save $$$ not having a DC-DC BUT, not even a fraction as much as just using a simple set of headphones!!
HOWEVER, he'd, of course, have to switch over from one battery to the other once he's down to 10.5V on the first batt.
He'd then be at 100 a/h (50 ea.) with a simple batt switch over in the middle somewhere.......But switching over is a pain? Esp when there's hot girls around!!
Nah.....easy peasy,
Of course you don't have to go down to 0 (zero) SOC in either case, that's just to get an idea of what we're dealing with here, you could switch over at 50% SOC, doesn't matter, the math would be the same wherever you chose to switch over.
A guy could also automate this switch over by using a pre-programmable low voltage cutout relay, but.......maybe some other day eh?
*
https://www.victronenergy.com/upload...50-400W-EN.pdf
** 2 hrs @ 42.5 A, 4 hrs @ 21.25 A, 8 hrs @ 10.63 A, etc, etc
