[Suggestion] Genesys AudioHook real-time contact center transcription (Node.js)

[deepgram-robot]

What to build

A working integration between Genesys Cloud's AudioHook protocol and Deepgram's streaming STT API, enabling real-time transcription of contact center calls within the Genesys Cloud platform.

Why this matters

Genesys Cloud is one of the largest enterprise contact center platforms, serving thousands of organizations globally. Their AudioHook protocol provides a standardized WebSocket-based interface for streaming call audio to external speech services. Contact center teams evaluating STT providers look for proven Genesys integrations — without one, Deepgram is invisible to this large enterprise buyer segment. A reference implementation removes weeks of integration work for contact center engineering teams and positions Deepgram as a first-class Genesys partner.

Suggested scope

• Language: Node.js / TypeScript
• Platform: Genesys Cloud AudioHook
• Deepgram APIs: Streaming STT (WebSocket), Audio Intelligence (sentiment, topics, entities)
• What to include:
  • AudioHook WebSocket server that receives Genesys call audio streams
  • Bidirectional audio routing: customer channel + agent channel with speaker labeling
  • Real-time transcription forwarded back to Genesys as transcript events
  • Audio Intelligence overlay: sentiment per utterance, topic detection, entity extraction
  • Health check endpoint and connection lifecycle management
  • Docker deployment configuration for Genesys Cloud integration setup
  • README with Genesys Cloud configuration walkthrough

Acceptance criteria

• Runnable with minimal setup (clone, add API key, configure Genesys webhook)
• README explains the Genesys AudioHook configuration clearly
• Uses current SDK version
• Handles both mono and stereo (dual-channel) audio streams
• Includes Audio Intelligence features (sentiment, topics) alongside transcription
• Deployable via Docker for production use

---

Raised by the DX intelligence system.