Overview
The Hecttor SDKs ship as plain libraries you can drop into any audio pipeline. For the platforms below we also publish a ready-made plugin, so you don't have to manage audio buffers yourself.
Supported platforms
- LiveKit: a
FrameProcessorplugin for LiveKit Agents, for Python and Node.js. Supports both ASR mode (Orpheus) and perceptual mode (Hermes). - Pipecat:
HecttorFilter, anaudio_in_filterfor Pipecat bots, wrapping Orpheus as a pre-ASR processor.
Both run entirely inside your agent process — all inference is local, and no audio leaves your machine.
Building your own integration
If you're integrating with a platform we don't list yet, the integration shape is always the same:
- Initialize the enhancer once at session start with your SDK key and config.
- Read the enhancer's configured chunk size after initialization and match your input chunking to it.
- On every chunk: call the per-chunk enhancement function and forward the enhanced output (or the original, if you're A/B-ing).
If you'd like us to publish a recipe for a platform that isn't listed here, let us know during onboarding.
Turn TakingSoon
Orpheus turn-taking detects when a speaker has finished so voice agents respond at the right moment, improving end-pointing and response timing. Coming soon.
Overview
The Hecttor LiveKit plugin adds real-time speech enhancement to a LiveKit Agents pipeline, cleaning participant audio before it reaches your STT and voice AI stack.