There is no reference to "TurnCoverage" in the available resources—however, TurnDetection is a key concept in realtime configuration. TurnDetection determines when a user has started or stopped speaking, typically using server-side voice activity detection (VAD). Configurable parameters include detection method, silence duration, activation threshold, prefix padding, and options for automatic response creation and interruption behavior. For example,

silence_duration_ms

controls how quickly turns are detected based on silence length, while

interrupt_response

manages whether the system can interrupt ongoing responses when new speech is detected. ActivityHandling manages how user activity, including interruptions, is processed. For instance, Google Gemini Live's

ActivityHandling.NO_INTERRUPTION

setting prevents the system from allowing interruptions, ensuring responses play out fully without being cut off. Both configurations help determine the flow of real-time conversations—TurnDetection handles speech boundaries, while ActivityHandling governs interruption rules. Let me know if you specifically meant "TurnCoverage" and need information on a different topic, or if you need code samples and usage scenarios for these features. Sources: Azure OpenAI Realtime API and LiveKit | LiveKit Docs | livekit-agents/livekit/agents/voice/agent.py | Turn detection and interruptions | LiveKit Docs