OpenAI Rolls Out Advanced Voice Assistant Following Scarlett

By Ben Sherry, Staff reporter @BenLucasSherry

OpenAI has begun rolling out its much-anticipated Advanced Voice Mode. It allows users to engage in human-like voice conversations with an AI assistant powered by GPT-4o, the AI unicorn's most advanced model. The new mode will gradually roll out to paid ChatGPT Plus subscribers over the next few months.

The Sam Altman-led company first unveiled the new voice mode in May during a press event, but much of the conversation following the event was focused on how one of the voice mode's vocal options, a female-sounding voice named Sky, had an uncanny similarity to the voice of actress Scarlett Johansson. Days later, Johansson revealed that Altman had asked her to be a voice of ChatGPT, but she had declined. After seeing a video of Sky in action, Johansson threatened legal action, and while OpenAI maintained that the similarity to Johansson was purely coincidental, it agreed to drop it as an option. 

An OpenAI spokesperson told Inc. that since the May preview, the company has been laser-focused on improving the vocal mode's safety, and has "made it so that ChatGPT cannot impersonate other people's voices, both individuals and public figures." The voice mode also won't be able to produce music or other copyrighted audio, so if you were hoping to use ChatGPT to generate a clip of Joe Biden singing "Espresso," prepare for disappointment. 

OpenAI says it will share a detailed report in early August on the recent work done to improve the voice mode. In the current version, users choose from a list of four voices, but have to wait for a response after every input. The new version, which OpenAI recently previewed and is now the one being rolled out, uses the same voices but is significantly faster and much more natural sounding. 

Photo: Getty Images.