parlot
Production voice AI observability — MIT OpenTelemetry instrumentation for LiveKit Agents, Pipecat, and LangGraph.
The Parlot Python SDK provides automated turn detection, audio metrics, and session contract tracking. Explore the reference documentation for the available modules below:
parlot.core
Shared telemetry conventions, base span processor, runtime context resolution, and session contract helpers:
base_parlotize, ParlotContext, ParlotRuntimeContext, SessionLogsCollector,
record_human_rep, set_session_metadata, and add_platform_ref.
parlot.instrumentation.livekit
Voice AI agent telemetry for LiveKit Agents: turn tracking, audio latency metrics, participant spans, SIP caller metadata, and audio stream egress.
View LiveKit Reference →parlot.instrumentation.pipecat
Voice AI agent telemetry for Pipecat: Conversation Contract turns and STT/LLM/TTS waterfall via OpenTelemetry remap.
View Pipecat Reference →parlot.instrumentation.langgraph
OpenTelemetry callback handler and automatic configuration for LangGraph and LangChain agent workflows — same turn model as voice AI agents.
View LangGraph Reference →Looking for guides and tutorials?
Visit the Parlot SDK Guides or start with the Quick Start Guide.