OpenAI is Cloning Human Voices

OpenAI is currently demonstrating its Voice Engine.

John Iovine
5 min readApr 1, 2024

AI has been used to create synthetic voices for several years. Numerous text-to-speech services are available on the Internet. In November 2023, Amazon’s Kindle launched a beta text-to-voice service for creating audiobooks from ebooks.

Recently, OpenAI demonstrated its synthetic voice program, Voice Engine, that can clone a human voice from a 15-second audio sample of the voice. This ability has the potential to be misused, such as deepfakes and unauthorized voice cloning. For this reason, the system is not being widely released. Instead, OpenAI is running small-scale testing to determine how to deploy this technology responsibly.

Scarlett Johansson Vs OpenAI

OpenAI released its new CgaptGPT-4Omni. This chatbox has a voice eerily like Scarlett Johansson’s. If you remember, Scarlett’s voice was used as the voice of an artificial personal assistant in the 2013 film “Her.”

OpenAI approached Scarlett Johansson to license her voice for this ChatGPT voice. Ms. Johansson turned down the licensing venture, so she was surprised when ChatGPT-4O was released, and the voice of “Sky” sounded very much like her voice.

--

--

John Iovine

Science writer, thinker, self-experimenter, focusing on personal development and health — www.john-iovine.com