Noveum Trace

Community-maintained observability integration that maps Pipecat pipeline events into Noveum traces and spans

Overview

NoveumPipecatTracer is an observability integration that maps Pipecat pipeline events into Noveum traces and spans — conversation turns, STT, LLM, TTS, tool/function calls, metrics, errors, and optional audio capture. Traces are viewable in the Noveum dashboard for debugging, analytics, and evaluation of voice agents.

Integration Docs

Noveum Trace documentation for the Pipecat integration

Source Repository

Source code, examples, and issues for the Noveum Trace SDK

PyPI Package

The noveum-trace package on PyPI

Noveum Platform

AI observability and evaluation platform for viewing traces

Installation

This is a community-maintained package distributed separately from pipecat-ai:

uv add "noveum-trace[pipecat]"

Prerequisites

You need a Noveum account and API key. Credentials are configured via noveum_trace.init() — the tracer itself does not accept api_key or project arguments. Set your API key in the NOVEUM_API_KEY environment variable or pass it to init().

Usage

Initialize Noveum Trace, then attach the tracer to your pipeline and task using the two-call API:

import os

import noveum_trace
from noveum_trace.integrations.pipecat import NoveumPipecatTracer

noveum_trace.init(
    project="my-voice-agent",
    api_key=os.environ["NOVEUM_API_KEY"],
)

tracer = NoveumPipecatTracer(record_audio=True)

pipeline = tracer.observe_pipeline(Pipeline([...]))
task = await tracer.register_task_handlers(task, transport=transport)

Once attached, conversation turns and per-service spans are exported to your Noveum project as the pipeline runs.

Audio and transcript capture are configurable: record_audio controls audio upload and capture_text controls transcript/text capture. Review these settings before using the integration with sensitive data.

Compatibility

Requires pipecat-ai >= 0.0.108 and Python 3.11+. See the source repository for the latest tested versions and changelog.