The respectful way is to approximate it using the phonemes available in the target language.
But that doesn't mean using the rules that language uses to translate the letters to phonemes! For example, English has a perfectly good sound for /i/ - so don't read foreign names with that sound in them as /ai/, just because the Latin spelling of said foreign name would be pronounced that way if it were an English word.
But that doesn't mean using the rules that language uses to translate the letters to phonemes! For example, English has a perfectly good sound for /i/ - so don't read foreign names with that sound in them as /ai/, just because the Latin spelling of said foreign name would be pronounced that way if it were an English word.