I haven't done this before, but I think you can accomplish it using one of the two following features:
Wwise Audio Input is a source you can use on a Sound SFX object that takes sound generated by the game at runtime. I think it's commonly used for VOIP chat and network-streamed audio.
Wwise External Source is instantiated in a Sound SFX object in the same way as Wwise Audio Input. This sounds like it works like Wwise Audio Input, except it points to an audio file that the programmer chooses.I think I would start with Wwise External Source. I bet your programmer can write the TTS to a .wav file as it's rendered in real-time, convert it to a .wem, and send it to your External Source object.