Okay, brainstorming on how I work with languages. Most of the time, since I'm doing one for a novel, I try to minimize stuff that would require me distinguishing between close sounds -- I STINK at telling vowels apart, so I barely use more than the Spanish/Japanese ah/eh/ee/oh/oo. (Once I even collapsed things to the Arabic ah/ee/oo). And, while the /r/ sounds might be trilled or not (or whatever), I usually don't care. I usually just want something that doesn't look too out of place when set into roman type. Most of the time I just use a consonant-vowel-consonant structure for my syllables, with maybe a couple of consonant pairs (like 'sm', 'tr', etc.). You could either pick out a bunch of consonants you like from a language (maybe use English's set, but omit a few -- then just vary the few between languages), and add in a few consonant pairs that exist in English and adjust the frequencies they occur.
About the only other time I check for pronunciation is near the end -- I have a name, and want to avoid it sounding 'wrong'. Here it's pretty arbitrary -- what sounds good to me might not sound good to someone else. (For example, someone mentioned being unable to adapt to a male name ending in 'a', since it's a feminine ending in some romance languages, which might seem silly to a Japanese speaker.) Since this isn't a hearing-specific problem, running it by other people probably should happen anyway. (Also helps to check that you don't accidentally name a character after a nerve gas (happened to a friend) or something similar to a foreign curse word.)
(For the record, I salute the amount of thought you're putting into this. Half of all published authors don't put this much thought into things. )