Besides copyright violations, a score is usually playing behind the vocals. Isolating the two would probably compromise the quality, if it's something any of us are capable of doing at all. So it just seems like a bad idea all around to me.
There is no possible way to do it accurately. I could isolate the music by re-composing it as a separate file. I would have no way to capture the dialog and use it.
Easy. Simply take a bunch of people's voices for a sample (females mainly) then adjust the pitch and lower the quality. Chances are we'll be able to make it sound *almost like the real Ducky. Lowering the quality will really help too. It'll make you think it only sounds a little different from Ducky because of the low quality. I'd say 2100hz and 48kps will work nicely.
That presents a few problems like, where to get voice samples, not to mention we would have to tweak every word that NPC says so they all sound like the correct NPC is saying them. In short, it would simply be too much effort and if we had to do this, we might as well not use voice acting at all.
Getting voice actors in general relies on finding, not just female voices, but female voices with really good microphones. I don't know if that will be an option around here.
The decision on voice acting will need to be made later. I'd like to have it, I just don't know how we'd pull it off yet.