One prompt → a full audio scene (speech, music, SFX).
Describe what you want to hear — the assistant writes a ready-to-use prompt.