for the DNM caps pls look here http://www.dnm.co.uk/acatalog/CapacitorShop.html
They've got the Slit foil with 1100uF 63V, so use two of them. But also the T-Network 10,000uF is great for the main "energy storeage" of Charilze. These caps are much better than standard types even as "better ones" with standard design approaches.
If you use battery supply so without any 50/60Hz ripple problem I suppose that even less capacitance is fine and other choices are availbale. So you could use smaller black gates, real foil capacitors (MKPs) to create 200 - 500uF. The sound of caps alter the sound very much of course and its worse trying around and get expirience.
Just check how much or less capacitance Charlize really needs... use small elcos of different value and check the dynamics. But just check the dynamics only as high frequency + resolution is altered also by diffenerent capacitor characters itself.
Test it first without cap at all
and then put in more capacitance to get a feeling for it. If proper prepared it takes two-three hours to get a rough estimation.
Of course measurements are fine too
but aren't possible without owning a costy oscilloscope.
If you end up keeping it simple and just use the caps "by hand" this test is still worth it. Often smaller cap values create a more faster "shoe tapping" - swinging sound with more micro dynmaics in place. While lots of capacitance kills high frequency extension and sounds heavy stomping in stead of swinging...
but I don't count in the DNM caps here this implies more for the normal constructions.
Its just my expirience and its probably not perfect science.