As I posted on the other thread I've had limited success by going back and working on it from a word to word basis (which is very very time consuming).

There will be instances where it may be necessary to go further and break into longer words and strip the sounds at either end and then apply a gate to the remaining sound componants and then reassemble them (and even then it will not be perfection).

If we has a team of people with no friends and no social life who could do this for each character we might get a workable dialogue bank by 2017, it really is a labour intensive job with no quick fix solution that I can see (other than breaking into Fort Skywalker and liberating the data tapes).

Are there any key scenes you need attempting?