Best workflow for narrated step-by-step physics explanations with the Genny API?

I am exploring a workflow that turns structured physics solutions into short, accessible voice explanations for students. The source material comes from Physics AI (https://physicsai.chat/), which produces step-by-step reasoning for typed questions and uploaded problem images.

For Genny API users, what is the most reliable way to map a multi-step solution into speech segments while preserving equation order, pauses between steps, and pronunciation of symbols such as delta, vectors, and subscripts?

Would you recommend generating one audio request per reasoning step and stitching the results, or sending the complete explanation with SSML-style pauses? I would also appreciate guidance on voice settings that work well for educational content without making equations sound rushed.