Hi Tony,
I honestly don’t know if this will help. You are correct that the phonetics that OpenEars will respond to in English approximately mirror the ones used in OpenEars’ English speech synthesis. What I can’t speak to is whether it would be a good user experience to hear what is probably going to be perceived as a name mispronunciation (in the cases where recognition isn’t good out of the box due to the name not being of English origin or relatively common) and be asked to use that mispronunciation by a software program for the benefit of the program. Names are sensitive subjects and the solution might feel kind of procrustean to the user. It seems like a case for some user testing.