In some cases characters with diacritics (the tilda and other extra markings, particularly relevant to rendering IPA) aren't available as a single character, but instead exist as the combination of two characters. Sometimes this works fine, other times not so much.
It may be worthwhile to play around with the html fonts you are using to see if the character rendering improves. You many also want to try different ways of processing from what I assume is originally an excel or text document, in regards to character encoding at that stage.
In short, I think this is a character encoding or font issue, not so much an ANKI specific issue. This doesn't help that much, I realize, but perhaps it helps narrow your focus.