An advanced audio mode for paying ChatGPT users is being rolled out, several months after OpenAI hit pause on the feature. The update brings five new voices (for a total of nine), improved accents, improved memory and more personalized interactions to the audio tool available for ChatGPT Plus and Team subscribers. Plans start at $20 a month.
The feature, originally announced in May, made headlines after one of the voice options sounded eerily similar to the fictional voice assistant portrayed by actress Scarlett Johansson in the film Her. After Johansson's legal team sent letters to OpenAI, the company delayed the greater rollout of advanced audio mode. In a blog post at the time, OpenAI said the specific voice, which the company called Sky, "was never intended to resemble (Johansson's)."
"Out of respect for Ms. Johansson, we have paused using Sky's voice in our products," the post said. "We are sorry to Ms. Johansson that we didn't communicate better."
The Sky voice option was removed from the feature.
The update promises the ability to better recall conversations and have more natural interactions. It also has the ability to converse in 50 languages and will listen to requests if interrupted.
"Hope you think it was worth the wait," CEO Sam Altman wrote in a tweet Tuesday.
Advanced audio mode is not yet available in all regions, including the European Union, UK, Switzerland, Iceland, Norway and Liechtenstein.
The launch comes as Google continues to roll out its free AI-powered Gemini Live audio tool to English-language Android users through the Gemini app. Gemini Live, which promises to learn and adapt to users' conversational style over time, was also announced in May. Meanwhile, Facebook-maker Meta announced on Wednesday it's adding celebrity voices, including those of Kristen Bell, John Cena and Dame Judi Dench, to its AI chatbot.