Audio Theater

Compose finished multi-character audio from a script or idea, fully offline. This skill is an orchestrator: it has no models of its own. It drives three local sibling skills and mixes their output with ffmpeg.

• voice-clone-narration -> one cloned/designed voice per character; one clean clip per line (these clips double as lip-sync reference).
• sound-effects -> foley / ambience cues (Stable Audio Open Small).
• bg-music -> instrumental score / beds / jingles (ACE-Step 1.5).