Melodyne will fix notes that are off, but if the smoking voice is harsh and raw, Melodyne will only fix so much.
timber in ones voice is best treated with EQ.
Vocal Synch will help glue them closer together however, what I have found is if they are way off, too much Vocal Synch can produce artifacts.
as for the levels, compress them both. then as mentioned previous, use volume automation.
if you need to be surgical, use nodes and draw them in. its almost like fader riding in the older days.
you can use Vocal rider but I find it just typically blends them all together and in some cases that's not what we want,
we want some vocals to sit in the back at certain times and others to stand up front, then vice versa.
the only way to accomplish this is through volume automation. which in my opinion...
is one of the top 5 tools a digital audio editor has.