Voice cloning

A voice clone is a model of someone's speech. The input is audio they really made: a social video, a voicemail, a school concert. The output is new words in that voice. It is synthetic audio pointed at a real identity.

Cloning is different from text-to-speech that uses a stock voice. A clone says a specific human is talking. Families should assume public videos of their children are enough material for a stranger to try.

The scam is simple and the voice is the proof people used to trust. A caller says your child is hurt, in jail, or traveling, and needs money now. The script rushes you. The defense is a family password and a callback on the real number. Do not argue with the voice.

Schools and teams post recitals and games. That audio is a gift to a cloner. You do not have to stop every recording. You should stop treating a familiar voice on an unknown call as identification.

HeyOtto's voice feature and speech-to-text and text-to-speech for Otto, the assistant. It does not clone a parent's or a child's voice, and it is not a product for making a real person say something new. Voice is available on paid accounts and HeyOtto is not voice-first. Kids type and read first.

Parents can still read the conversation. If a cloned voice reaches you from outside HeyOtto, hang up and call back on a known number, and report the scam at ic3.gov.