Voice cloning creates a synthetic voice from a sample recording, which can then read out any text, often even in a different language from the one the original recording was in. Companies use it for dubbing videos during expansion, audio versions of articles, voice announcements and narration in e-learning. Legally and ethically it is a sensitive area: using a specific person's voice requires their demonstrable consent, and the scope of use should be defined contractually, including its period of validity. The listener needs to be informed that the voice is synthetic if they could otherwise be misled. Technically, quality today is high for neutral delivery, but it still lags behind for emotion and specialist terminology, and the output needs checking before publication.
See also: AI for Video Subtitling, Disclosure of AI interaction, Localisation vs translation.
