This may or may not work because I would have to have the tracks here to make actual judgement calls. That said, IN GENERAL.....
I assume they are all singing or rapping at the same time, if there is a main vocalist and backup singers the following method would not apply and I would take another approach.
I would put the music on 1 master bus in your software and the all three voices on master buss 2. Do a rough mix by ear first to get levels about where they need to be. Put the SA on the vocal mix of all three voices. See what the spread (frequency dominated area) is on the voices.
Because the music is already mastered I would only do a slight db reduction across the vocal range in the music track. On a parametric this would be a Q of 0 with a db reduction of 2-4 centered right in the middle of the vocal range.
Check the mix again, if there is some dominant voice or voice section in the music that needs to be pop out in the mix, you might go searching for those frequencies that pop the voice and do some narrow band pushes of around 3-4 db - Q of maybe 20-30 for just those frequencies.
This is vocal carving but there are other approaches, such as pan the music tracks slightly left and right leaving a hole for the voice, or even ducking the music in certain frequency bands to make them softer when the vocal comes in. You can also explore the one mentioned previously, push the music into reverb a bit when the voices come in. Because this is a music track already done, I assume the lo end is already happening, I probably would not go down the reverb path.
I am sure there are others out there who are far more experienced than me and have their own special tricks for carving out a hole for the vocal.
Regarding building the music, this means the usual starting technique of working the drum kick and bass at the low end but be aware of when you are crossing into a baritone, like Knopfler, and thus putting hard shelves on these instruments. His voice has some rich qualities at 160 so I will lo pass at 160 and below. Just a small shelve of about 4 db. Then I would pick up some of the hi end bass sound on the other side of the voice above 1.25 with some narrow Q pushes on important tones. This is the snappy sound of the bass. The same would hold true for the kick, and guitars and piano. Get what you can from the lo end without walking on the voice. Actually you can spend a lot of time here at the lo end getting the music to work.
Above the voice you take the opposite approach with some hi pass above 1.25 for musical instruments. Build a nice mix of instruments on the top side.
Now you have created a nice space for the voice. These frequency numbers would change for every vocalist by the way. Please bear in mind this is only one approach, sometimes its best to slap an Art Tube on the stereo buss and just fuzz it up to get a blended sound. Hey its art, you can do what you want.
