I’m very interested to see if/how the model can figure out to produce a different voice when a character is speaking and how to keep the same voice for each cha

I’m very interested to see if/how the model can figure out to produce a different voice when a character is speaking and how to keep the same voice for each character across the whole book consistent. Especially the second problem is not trivial at all from my understanding of how neural networks work.