Pipecat Integration
The Hecttor Pipecat integration brings real-time speech enhancement into a Pipecat voice agent pipeline. Incoming user audio is cleaned by the Hecttor SDK before it reaches VAD and your STT service, which improves transcription accuracy in noisy environments without you having to manage audio buffers yourself.
The integration is a Python package, pipecat-hecttor, published on PyPI and listed in Pipecat's Supported Services as a community integration. Unlike the LiveKit plugin, you install it by package name — only the Hecttor SDK itself is a licensed download.
When to use it
- You run a Pipecat bot and want cleaner audio into your STT / LLM pipeline.
- You want the enhancement to run inside the bot process, with no extra network hop — all inference is local, and no audio leaves your machine.
- You prefer a drop-in
audio_in_filterover wiring the SDK into your own audio loop.
The integration works with any Pipecat transport that supports input audio filters (Daily, LiveKit, WebSocket/Twilio, local WebRTC, ...): processing happens in the bot process, so the transport does not matter.
If you are not on Pipecat, use the direct SDK integration instead — see Orpheus Getting Started.
How it works
The package provides HecttorFilter, an implementation of Pipecat's BaseAudioFilter interface — the same slot used by Pipecat's first-party noise filters — passed to any transport via the audio_in_filter parameter:
- The transport receives the user's audio and hands each frame to the filter before VAD and STT.
- The filter buffers audio into the model's chunks (20 ms by default) and passes each chunk to the Hecttor enhancer, which denoises it in real time. Added latency is bounded by one chunk.
- The enhanced audio continues through your normal bot pipeline (VAD, STT, LLM, TTS).
The filter wraps Orpheus (ASRSpeechEnhancer — models optimized for machine transcription, the right choice in front of STT), using a voice-isolation model by default, which suppresses background speakers as well as noise.
Requirements
| Requirement | Value |
|---|---|
| Pipecat | pipecat-ai ≥ 1.5 |
| Python | 3.11+ (Pipecat's floor) |
| Hecttor SDK | Python SDK ≥ 3.1.0, installed alongside the package |
| Platforms | Same as the Hecttor SDK: Linux x64 / arm64 (glibc 2.28+), macOS, Windows x64. The pipecat-hecttor package itself is platform-independent. |
| Hecttor SDK key | Required — issued during onboarding with the Voice AI Agent Enhancement type |
Next steps
Continue to Getting Started for installation and the integration steps.