10uf Wima should be fine, I personally wouldn't use an oscon for signal but feel free to try it

The standard output stage +/- should be mirrored, it isn't for some reason and this gives the DC offset.
If you look at resistors R8 right channel and R9 left channel for the inverted side they are 36k, the mirrored side to these for the non inverted R14 right channel and R17 left channel are 22k1.
Basically R8 and R9 should be 22k1 to match 22k1 for R14 and R17.
If you change those resistors to match, and then measure the output pins of the op-amp (pins1 and 7 to ground) the DC offset should be a couple of mV's instead of the hundreds of mV's in the standard circuit.
You may lose a bit of gain but it shouldn't be a problem.
BTW a thanks to Ray for pointing this out

theres also a cap across each of those resistors, they don't match either only thing here is that on the non inverted side they are 1N and the inverted side they are 56p, I'm not sure if having 1NF across the NFB resistor would sound worse than 56p, its something I'd need to play about with, these caps shouldn't effect the DC offset though, its mainly the resistors mentioned above.

This dac is weird

I'd still use a input coupling cap in the amp or pre just as a precaution