Comment by jkoff

I first generate a script to be handed to a TTS model. For this step, Claude 3.5 Sonnet works well. For voice synthesis, I've been using Google Cloud's Text-to-Speech API and it's been adequate.