ProductGPT Live / GPT-Live实时语音
GPT-Live
A real-time voice interaction system from OpenAI built on a full-duplex audio processing architecture. It supports simultaneous interpretation, parallel task execution, and paralinguistic signal simulation, enabling an AI Agent capability in which foreground conversation and background task processing occur concurrently.
Timeline (last 90 days)
Oct 13
Higgsfield describes GPT-Live as a path to "voice filmmaking."
Unverified50%
Sep 12
Devin Voice's capabilities are supported by two core technologies, GPT-Live and SWE-2, responsible for voice understanding and code generation respectively.
Unverified50%
All Facts (2)
Source Articles
GPT-Live-1.1 Upgrade: New Voices and Accent Improvements TestedOct 15Higgsfield Teams Up with GPT-6 Astra: AI Video Length Breaks the 10-Minute BarrierOct 13GPT-Live Real-Time Voice Interaction: How "Listen and Speak Simultaneously" Enables Natural ConversationJul 20OpenAI GPT-Live: The Full-Duplex Voice AI RevolutionJul 20How GPT-Live Solves the Cocktail Party Problem in Voice AIJul 19GPT-Live Hands-On: Voice Finally Gets GPT-5.5 — How Good Is the Full-Duplex Experience?Jul 15ChatGPT's New Voice Feature Goes Live: A Deep Dive into the GPT-Live Full-Duplex ExperienceJul 15OpenAI GPT-Live Full-Duplex Voice Model: Reshaping the Human-Computer Interaction GatewayJul 12