Pipecat

What it is

A reference integration showing how to use Orpheus as a pre-ASR audio processor inside a Pipecat pipeline. Orpheus sits between the audio input and the ASR service in your Pipecat graph, cleaning audio before transcription.

Who it's for

Teams building voice agents on Pipecat that want to lower WER on their ASR step without changing their ASR provider.

What we'll publish here

• A Pipecat FrameProcessor wrapper for Orpheus.
• Configuration that matches Pipecat's default audio shape.
• A runnable example showing Orpheus → ASR → LLM in a Pipecat pipeline.
• Notes on chunk alignment between Pipecat's audio frames and Orpheus's chunk size.

Until then

The integration shape is straightforward: implement a Pipecat FrameProcessor that buffers incoming audio frames to Orpheus's configured chunk size, runs them through Orpheus's per-chunk enhancement, and emits the enhanced frames downstream to your ASR processor. See Orpheus Examples for the chunk shape, and ask us during onboarding if you'd like a code review on your draft.