Loanwords are pronounced exactly the way they are transcribed. Depending on the circumstances of the transcription (which are often unknown), the transcription is based on a mix of actual pronunciation, its alphabet representation and the weather.
If "Melbourne" becomes メルボルン, then it will be pronounced exactly like //meruboruɴ//, no matter whether it's far from or close to the original pronunciation. (It could also have been メルボーン.)
The insertion of the extra vowels follows some loose rules: the default is u. However since t+u becomes [[tsu]], after t you usually use o instead of u, as in アート āto "art". However this may be broken if tsu is perceived to be a similarly good fit, e.g. ピーナツ pīnatsu "peanut" or "peanuts". Sometimes the choice is also just taken to be the same as the vowel that follows it: for example in a syllable with two consonants and the vowel i, the default u that would be "inserted" after the first consonant may also become i, to avoid the contrast –u–i, as in キリスト kirisuto "Christ".
There have been efforts to include phonemes from other languages, such as //va vi vu ve vo//, transcribed as ヴァ ヴィ ヴ ヴェ ヴォ, but by many still pronounced [[ba bi bu be bo]]. (For example, ヴィオラ vs. ビオラ. Also see Do Japanese actually pronounce the "v" sound?)
The introduction of スィ ([[si]] rather than [[ɕi]]) or ディ ([[di]] rather than [[dʑi]]) has been successful. For example, older sources use エジンバラ Ejinbara for Edinburgh, but エディンバラ Edinbara now seems to be standard and most people do pronounce ディ as [[di]].
Having to introduce extra vowels into loanwords, like Meruborun can lead to curious effects when actually pronounced. For example プロ from "pro(fessional)" has pitch accent プロ【HL】, which sounds like the first vowel in puro (which shouldn't be there) is stressed.
So, unlike in English (or French), in Japanese everything that is written is pronounced — sometimes even with a pitch accent that sounds like vowels which weren't present in the original word are stressed.