I've actually done it on l/r panned comped vocals for stereo effect. Did it take a long time? Fuck yeah, but it is possible. In that song, I had about 30 takes of the same chorus, which was really only 30 seconds or so long. I cut out the best parts of each take, until I had two viable comps, then pitch corrected (which can require further slicing the vocal in melodyne to correct small tonic variations), and then time stretched the various slices to match what I thought was the best take rhythm wise, which I used as a guide for the time. Of course, any defects caused by the adjustments were hidden behind some reverb, rapid fades and cross-fades, and the backing track generally. Total time for 30 seconds of tape? Probably 20 hours. 😒