Reply To: OpenEars and the main thread

[hohl]

Yes, the last approach sounds like the best solution. I just have to look up how the LanguageModelGenerator works and how to use it with already pre-created pronunciation.