The Audio tab lists twelve built-in voice presets — low and weathered, clipped and procedural, warm and close, bright and young, RP precise, Scots low, Southern slow, documentary narrator, child, elder, synthetic flat, broadcast urgent.
Each carries accent, pitch, pace and emotional range. These describe a voice in terms a TTS adapter can act on; they are not recordings of real people, so no consent record is needed.
A cloned voice is a different object entirely. Cloning a real person's voice requires a verified consent record with identity verification. Until consent is verified, assigning that voice to a character is refused — not warned about, refused. This is enforced in the service, so it cannot be bypassed by the UI.
Cast a voice per character from the Audio tab. Fourteen emotions are available for dialogue direction.