Pipecat

Trace Pipecat real-time voice and multimodal AI agents in Muster via OpenTelemetry.

Pipecat is an open-source Python framework for building real-time voice and multimodal conversational AI agents. Muster captures conversations, turns, and per-service calls (STT, LLM, TTS) via OpenTelemetry.

Setup

1. Get Muster API keys + base64-encode them

Auth headers for OTLP use Basic auth:

echo -n "pk-lf-...:sk-lf-..." | base64

2. Configure environment

ENABLE_TRACING=true
OTEL_EXPORTER_OTLP_ENDPOINT="https://app.getmuster.io/api/public/otel"
OTEL_EXPORTER_OTLP_HEADERS="Authorization=Basic%20<base64_encoded_keys>,x-langfuse-ingestion-version=4"

3. Initialize tracing

from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
from pipecat.pipeline.task import PipelineTask, PipelineParams
from pipecat.observers.tracing import setup_tracing

exporter = OTLPSpanExporter()
setup_tracing(service_name="pipecat-demo", exporter=exporter)

task = PipelineTask(
    pipeline,
    params=PipelineParams(enable_metrics=True),
    enable_tracing=True,
    conversation_id="customer-123",
)

Trace structure

Conversations are organised hierarchically:

A conversation trace contains multiple turn spans.
Each turn contains spans for STT, LLM, and TTS service calls.
Performance metrics, token usage, and character counts are attached.