What is voice cloning?
Voice cloning creates a synthetic voice from recordings of a real person’s voice. The result can then speak new text that the person never recorded.
Why it matters
A cloned voice lets an organisation keep one recognisable voice across its phone menus, announcements and voice agents, and update wording without new recording sessions. Because the voice belongs to a real person, it also carries risk. The consent of the person whose voice is used is required, and the organisation should be clear about where and for how long the voice may be used.
How it works
The person records a set of sample speech, usually by reading prepared scripts in a quiet room. A model analyses those recordings and learns the features that make the voice distinctive, such as pitch, pace and tone. That learned voice is then attached to a text-to-speech system. From then on, any text given to the system is spoken in a voice that resembles the original speaker.
Example
A bank has used the same voice artist on its phone line for years. With her written consent and an agreement covering how the voice may be used, she records a set of scripted sentences. The bank’s voice agent then uses the resulting synthetic voice, so callers hear the familiar voice whether a sentence was recorded in advance or generated during the call.
