ChatGPT’s upgraded voice mode is better at shutting up

OpenAI has launched GPT-Live-1, a sophisticated voice model designed to mimic natural human conversation by enabling full-duplex communication. This allows the AI to speak and listen simultaneously, wait during pauses, and be interrupted by the user, representing a significant upgrade over previous turn-based models. Key features include real-time translation, the ability to display AI-generated visuals for data like sports scores or weather, and improved interaction etiquette.

To ensure safety, the model incorporates built-in safeguards, including protocols for crisis helpline support and age-appropriate content for teens. While GPT-Live-1 is available to Go, Plus, and Pro subscribers, OpenAI is also providing a more efficient version, GPT-Live-mini-1, for free users across iOS, Android, and web platforms.

(Source：The Verge)