Voice & smart devices
Explore more natural human–machine interaction.
One sentence connects a person's intent with a coordinated response from devices.
Home, office and in-car devices each have their own controls, so users must remember different commands and logic. With follow-up commands, vague requests or tasks that span devices, traditional interaction struggles to understand context, and the experience across devices feels disjointed.
Yuanzhu uses Yuan MetaCore-K3 for natural language understanding and contextual reasoning, organizes intelligent applications for voice interaction and device control through YZ MIX-5A, and draws on the task planning and multi-agent collaboration of Yuan Agent-Turbo to explore interaction that runs from user intent to multi-step execution.
Once speech recognition, speech synthesis, device interfaces and scene rules are connected, the system interprets user commands, works out the task steps and sends the right requests to the right devices. For door locks, vehicles, industrial equipment and other high-risk operations, the system applies identity verification, permission checks and a second confirmation.
Users express what they need in more natural language, and the system responds based on context and device capabilities. For connected and authorized devices, several steps can be completed together under preset rules, making interaction more fluid.
One scenario, end to end
- Voice inputCapture the user's command through speech recognition, handling background noise and voice input.
- Understand intentYuan MetaCore-K3 helps interpret natural language, context and the task goal, clarifying what the user wants done.
- Plan and dispatchYuan Agent-Turbo organizes multi-step tasks and arranges execution through the YZ MIX-5A entry point and device interfaces.
- Device response and feedbackCall device capabilities within permission and safety rules, and report results by voice or other channels.