pronunciations do not work as expected

If I pass in pronunciations via the API it doesn't actually render with the expected target pronunciation unless I first go into a a voice project in the UI and add it to the pronunciation library.

In my opinion this kind of defeats the purpose of the API and should be fixed.