A week after releasing its latest multimodal AI model, GPT-4o (“o” for “omni”), OpenAI is now pausing the use of the infamous female voice dubbed Sky in ChatGPT following criticism.
The OpenAI demos showed the multimodal AI’s capabilities which included reading emotions, performing live translations, and even solving Maths and coding problems in real time. The new voice capabilities would allow users to have more humanlike conversations with the AI chatbot. However, critics found Sky’s voice “too flirty” and overly sexualized.
Sky’s voice sounded uncannily familiar with Scarlett Johansson’s voice in the 2013 film “Her.” The film is about a lonely writer who falls in love and becomes more dependent on an AI voice called Samantha (played by Johansson) as he copes with the aftermath of a failing marriage.
After the announcement of GPT-4o last Monday, OpenAI’s CEO, Sam Altman posted the word “Her”, referencing Samantha (played by Scarlet Johansson) in the film.
OpenAI has since denied claims that Sky’s voice was inspired by Johansson’s AI voice in Her.
“We believe that AI voices should not deliberately mimic a distinctive voice – Sky’s voice is not an imitation of Scarlett Johansson but belongs to a different professional actress using her natural speaking voice. To protect their privacy, we cannot share the names of our voice talents,” OpenAI has shared in a release.
The AI company has also explained how the various ChatGPT voices were chosen. OpenAI selected 5 voices out of 400 submissions received. Users on ChatGPT’s free plan can select among the five distinct voices: Breeze, Cove, Ember, Juniper, and Sky – currently not available – to interact with the AI chatbot.
OpenAI says more voices are coming to ChatGPT plus users “to match the diverse interests and preferences of users.”
