depends on the genre and the vocal performance... generally i use a very high ratio (like 10:1 at least) and then adjust the threshold to taste. this is with a normalized audio file too. eq is also never really set in stone but i like to scoop low mids and cut extreme lows and highs in nearly all cases, specific frequencies depend on a lot of things! where you put the eq in the chain is important too, you might want to place an eq before to make the comp a little less reactive and then an eq after to shape the vocal tone more.
mind you i pretty much never use multiband comps on vocals and instead opt for a de-ess & compressor combo with eq in between. the more tracks of vocals i use for one part the less i try to process them.