[Suggestion] Vonage real-time call transcription with streaming STT (Node.js)

[deepgram-robot]

What to build

A Node.js application that captures live audio from Vonage voice calls and transcribes it in real-time using Deepgram's streaming STT API. Demonstrates the complete telephony-to-transcript pipeline using Vonage's WebSocket audio streaming.

Why this matters

Vonage is a major CPaaS provider used by thousands of businesses for voice communications. Developers building call center analytics, real-time agent assist, or compliance monitoring need a working example showing how to stream call audio to Deepgram. This complements existing Twilio integration examples and gives developers choice across telephony providers — a common requirement for enterprises evaluating CPaaS options.

Suggested scope

• Language/Framework: Node.js / Express
• Deepgram APIs: Streaming STT (Nova-3) via WebSocket
• Vonage APIs: Voice API with WebSocket audio streaming
• Features: Inbound/outbound call → Vonage WebSocket stream → Deepgram live transcription → real-time transcript output
• Complexity: Medium — requires Vonage account, webhook handling, WebSocket piping
• Backend required — Express server with ngrok for local development

Acceptance criteria

• Runnable with minimal setup (clone, add API keys, configure Vonage app, run)
• README explains the pattern clearly with architecture diagram
• Uses current SDK versions
• Demonstrates real-time transcript output during an active call
• Handles call lifecycle events (answer, stream, hangup)

---

Raised by the DX intelligence system.