Overview
Enable natural voice commands and interactions between a robot and the people around it. With the LiveKit Agents framework, you can add high-level voice control to any robot, so an operator or bystander can direct it in plain language.
Common patterns
Voice apps commonly combine several capabilities:
- Voice commands: Tool calls from the agent trigger robot actions over RPC.
- Conversational feedback: The robot responds and reports status in natural language.
- Long-running actions: Run a robot action in the background using async tools so the agent keeps talking while the work completes.
- Multiple speakers: Use speaker diarization to tell speakers apart when several people talk to the robot, and focus on the primary speaker while ignoring background chatter.
Next steps
For implementation details, follow the Agents guides: