OpenAI Launches ChatGPT Voice on Desktop with Computer Control and GPT-Live
OpenAI has launched ChatGPT Voice for its desktop application, allowing users to control their computers and direct multiple agents using voice commands. This feature is powered by the new GPT-Live model, which supports simultaneous speaking, listening, and task coordination. This update marks a significant step forward in hands-free human-computer interaction, enabling users to manage complex workflows across different agents using natural speech. It showcases the practical integration of real-time, full-duplex voice models into daily productivity environments. The feature is powered by GPT-Live, which utilizes a full-duplex architecture to enable natural, conversational flow. Users can direct agents running in ChatGPT Work or Codex, and the update is rolling out globally.
## BACKGROUND
Traditionally, voice assistants operated on a half-duplex basis, requiring users to wait for the AI to finish speaking before responding. OpenAI's GPT-Live introduces a full-duplex architecture, allowing the AI to listen and speak simultaneously, mimicking real human conversation. Additionally, AI agents are autonomous programs designed to perform specific tasks, and coordinating them has typically required manual text inputs.