You know, if you want to be technical... the ch in Melchior should be prounced not as anyone has yet said it, not like a k, but rather as the ch you might get in the German nacht or, more accurately, a Semitic sound. It is, after all, a Semitic word. I can think at least one other name using the Melch- element, that being Melchizedek. If I'm recalling this correctly, it has the meaning of 'king'. The 'i' is an 'of' or something as that, and in this name at least zedek is 'righteousness.' However, I'm not certain what the 'or' is in Melchior. Anyway, the point is... though we colloquially prounce it (or you should) Mel-key-or (most natural for English), it's not capturing the sound of the actual name.
The same might be said for Janus as well. The J is in fact not a j as we have it, but rather a consonantal 'i'. As such, it's an I sound. Like in the saying 'alaea jacta est.' It's not 'alaya jakta est', rather it's 'al-AI-a YAK-ta est.'
Magus... might have a dj sound for the g in its distant Persian origin, but as it stands, it's been in the Greek, then Latin, vocabulary for the last 2500 years, and as such the standard pronounciation in either of those languages which lack just a sound is a hard g. Nor does the a make a nasal 'a' sound as some seem to have it (this 'MAY-guss'), but rather... well, honestly, I'm not sure of the term of pronuncation. I'm a philologist, but more on etymology and less on pronounciation (what I get for studying unspoken languages.) As it is, it should be 'Mah-guss', technically.
As for those pronouncing it KRAW-no, well, that might be a little closer to the actual pronounciation. Since it comes from khronos, that o is short, and not like our long o that we're wong to put into that position.
The thing is, since many of these names are drawn from older names, of either IE or Semitic origin most of the time, we have to somewhat draw a distinction between the colloquial English pronounciation and the more technical old one, and make a choice between these.
