OpenAI Pauses Sky Voice in ChatGPT Amid Concerns Over Similarity to Scarlett Johansson


OpenAI is temporarily halting the use of the Sky voice in ChatGPT due to concerns that it closely resembles the voice of Scarlett Johansson, known for her role in the film "Her."

According to OpenAI, the voices in ChatGPT are provided by paid voice actors. From an initial pool of 400 candidates, five final voices were selected, and it is purely coincidental that the actress behind the Sky voice sounds similar to Johansson.

Voice interaction is set to become a key feature for OpenAI with the introduction of the new GPT-4o model in ChatGPT. This model will bring an advanced conversational interface, allowing users to engage in real-time conversations with a natural-sounding, emotionally responsive AI.

The resemblance of the Sky voice to Johansson's voice gained attention when OpenAI CEO Sam Altman and others noted the similarity between the new AI model and Johansson's character in "Her." In the movie, Johansson voices an AI operating system named Samantha, who forms a romantic relationship with a lonely writer played by Joaquin Phoenix. The emotional depth of GPT-4o has drawn clear parallels to this depiction.

What sets GPT-4o apart from previous models, including earlier versions of ChatGPT Voice, is its multimodal capability. It has been trained to understand and generate images, text, video, and speech, enabling real-time, emotionally nuanced conversations.

Due to the enhanced emotional expression in ChatGPT Voice, there is a potential risk of misuse, such as creating deepfakes. This concern is understandable, especially from Johansson's team, given the striking similarity between her voice and the Sky voice.

Currently, ChatGPT Voice offers five voices: Breeze, Cove, Ember, Juniper, and Sky. These voices will also be available at the launch of the new version. OpenAI partnered with selected voice actors, licensed their voices, and used samples to train the AI voice models. The company stated in a blog post: "Each actor receives compensation above top-of-market rates, and this will continue for as long as their voices are used in our products."

OpenAI began searching for voice actors early last year, using award-winning casting directors. Out of over 400 submissions, five actors were chosen and flown to San Francisco for recording sessions. These sessions provided the samples to train the new AI voice models, with each actor corresponding to one of the five voices.

“We believe that AI voices should not deliberately mimic a celebrity's distinctive voice. Sky’s voice is not an imitation of Scarlett Johansson but belongs to a different professional actress using her own natural speaking voice,” the company said, noting that they could not disclose her name.


Popular Posts