OpenAI Rolling Out More Natural Advanced Voice Mode for ChatGPT

OpenAI has begun rolling out Advanced Voice Mode to a select group of paid ChatGPT users for testing. This new mode enables more natural, real-time conversations, allowing interruptions and the ability to detect and respond to humor and sarcasm. Unlike the current voice mode, Advanced Voice Mode doesn’t require converting speech to text and back, resulting in lower latency. OpenAI demonstrated this feature in May with an AI voice named Sky, which sounded similar to Scarlett Johansson. Johansson, who had declined offers to be ChatGPT's voice, was upset by the resemblance and took legal action, leading to the voice's removal. OpenAI stated that Sky wasn’t intended to mimic Johansson and has since focused on improving voice interaction safety and quality. The new mode now uses four preset voices and blocks attempts to create unauthorized or violent content. Selected users will receive emails with access instructions, with a broader rollout planned for the fall. OpenAI aims to enhance the feature through early user feedback before a wider release.