We need the ability to define words in advance & hear how they are pronounced so that we can phonetically correct the words and build a common vocabulary that all podcasts can rely upon. For example, “ROI” is mispronounced as a singular word versus a clear and distinctive “R O I” - so being able to build out a pronunciation so that the system knows how to correctly pronounce “ROI” when it sees the word is critical. The same could be true for names, etc.
