Only one thing I can think of without telling you to redo everything, and this is gonna be dicey. I don't if the results will actually be any good.
The fastest way will be to re-import the music file (as a second file, different file name) and allow SRC to properly convert it. You should then have both your buggered music file and the new version that is at the correct speed and pitch.
First consolidate your vocal track into one file that is the exact length of the music region it was recorded against. Find the exact length of the new (correct) region (you may want to view in sample), and varispeed time-compress the incorrect music file to match. Play back and A/B between the two music regions to make sure that it worked properly. Now, time-compress the vocal track (NO VARISPEED this time) to match the exact length of your music files, and you should be there.
Again, I don't know how good the results will be...but it's worth a shot.