OpenAI releases ChatGPT’s hyper-realistic voice to some paying users

OpenAI introduced GPT-4o's Advanced Voice Mode to a limited group of ChatGPT Plus users with hyper-realistic audio responses.

: OpenAI has started rolling out GPT-4o’s Advanced Voice Mode to some ChatGPT Plus users. This feature offers hyper-realistic human-like audio responses. The full rollout is expected by fall 2024, excluding video and screensharing for now. More safety filters and preset voices are included to avoid controversies.

OpenAI began deploying GPT-4o’s Advanced Voice Mode to select ChatGPT Plus users on Tuesday, allowing them to experience hyper-realistic audio responses. Initially accessible to a small group, the full rollout is planned for fall 2024, but video and screensharing features are not yet included and will launch later.

In May, the feature stunned audiences with its real-human resemblance, notably sounding like actress Scarlett Johansson. Despite OpenAI denying the use of Johansson’s voice, legal complications led to the removal of the voice from the demo, delaying the Advanced Voice Mode’s release to enhance safety measures.

The new voice mode uses GPT-4o without auxiliary models, promising lower latency with emotional intonation recognition. To avoid deepfake issues, OpenAI has limited the feature to four preset voices and included new filters to block unauthorized requests, addressing potential legal concerns surrounding voice impersonation and copyrighted audio.

deepfakesGPT4oAI technology