Multilingual TTS mispronounces brand names differently in every language. How do you solve this?
I run a video localization SaaS where users generate their own voiceovers/dubs in multiple languages with AI voices.
Problem: a brand name is spelled the same everywhere, but each language's voice pronounces it differently, so it sounds wrong/inconsistent across locales. I can't hand-fix each name, because my users generate the content themselves and telling them to write phoneme tags isn't realistic.
For those building on top of TTS: how do you keep brand/product names consistent across languages when your users drive the generation, not you?