For years I've been following the common and age-old advice for vocal compression, which is to start at about a 3:1 ratio, using a fast attack time (like 0-to-10 millseconds so little-to-none of the vocal is heard above the threshold), and medium slow release time (50-to-80 milliseconds so the notes decay naturally). Then set the threshold to -20 dB. You'll then need to adjust the threshold setting until only the loudest notes are being reduced, and that reduction should not be much more than about 3 dB. Then just use your ears to make adjustments until you get things just right. To review what the heck I'm talking about with all these settings, see my video post Vocal Compression Using Reaper’s ReaComp Effect .
But as with all things audio, you should use your ears to determine what is "right." Still though, it's hard to stray too far from what you've been taught is the "normal" way of doing things. I just re-watched a video (see my post An Interesting Vocal Compression Tip) that I wrote about here a few months ago that suggests a bit of a different approach to vocal compression. At the time, I sort of thought it was interesting and suggested you might want to try it out. However, I did not take my own advice! Now I have, though. And I have a new normal.
In the video, Andrew Koss explains his technique of using super fast attack AND release times (like zero on both), with a very low ratio (try 1.5:1 or 2:1), and gain reduction of -10 or -12 dB. I tried it and really like the way it makes my vocals sound.
One thing to be careful of, though, is sibilance (that enhanced SSSSSS sound) being hyped, which can happen on some compressors (or compressor plugins) when extreme settings are selected. If you REALLY like the added pop in the voice that this technique gives you, and you get a bit of excess sibilance, you can always use EQ to reduce it. See our article - How to Fix SSS-Sibilance in Your Audio With Sound Editing Software.
Try it and see what you think!