I am mixing a vocal track of my friend performed on a karaoke track.
The karaoke track itself seems mixed and mastered.
So:
1. Should I master the whole mix again?
2. How to glue the vocal to the karaoke?
Compression? Oh yes! but what should be my starting points? Guide me here!
I always overcompress my louder parts in the vocal!

 

1 in this case you should unamaster the track. Nowadays There are som many unverb, unchirps and só on. Aliso use an expander To undu the compression of the track.
2 use parallel compression, much more than only one comp, multiband compression like the excelent from Vertigo. Try to do a lot of parallel fx like dirty channel, dirtier and cleaner, and diferente ambiences like close, stage and ambience. Use channel strip emulatios and different tape and valve emulators, and lots of soft distortions. Use a little of the fx from the voice for the new raw base track.
And mainly search! Here thers a lot of excent tutorials and in the web you could spend a life learnig right?

 

Naw, just carve out some 500-ish Hz in the mix, eq/compress/limit/reverberate vocals like regular and then make some final adjustments in the master. No need to go crazy on it.

 

What I would try 1st: Run your stereo music track through stereo graphic and tweak it and your vocal track with it's own graphic together. If you still hear something jumping out, add a small amount of light compression--- just enough to take out any major spikes or holes in the over all blend. If whole sections of the song need adjusted in relative volumes, use an envelope on that section, or automate the mix. If your friend wants something more serious than that, they should get away from K-Row-Wacky.