Do you think you could add custom phonetic overrides for words in the future? I can think of a few semi-simple ways to implement it. Some premium voices have SSML phonetic support, like Polly. Wavenet voices (which I prefer overall) you can fake phonetics by simply feeding in words with ' between them. For example, if the name is Hakuro, wavenet pronounces is Hak or o. But if you plug in Hak'uro, it changes the emphasis and pronounces it Hak your o, which is correct. So even if you just started with a super simple implementation of replacing specific words with user defined versions, it would be 85% of the way there, and fully customizable by users based on their chosen voice. Consider, we could even utilize this to replace mass ellipses which some voices, even wavenet, struggle with. "......" is often read as "ellipse, ellipses, ellipses, ellipses, ellipses, ellipses" But we could just replace "..." with "," and it would pause properly.