Queue: Sinch Voice API call transcription with Deepgram STT

[github-actions]

Integration: Sinch Voice API

Origin

Requested in #142: "Server-side Node.js using the Sinch SDK (@sinch/sdk-core). Handles inbound PSTN/SIP calls via the Sinch Voice API, streams audio via the SVAML conference bridge or WebSocket audio hook, and forwards to Deepgram live WebSocket."

What this should show

A Node.js server that uses the Sinch Voice SDK (@sinch/sdk-core) to handle inbound PSTN/SIP calls, tap the call audio via the SVAML conference bridge or WebSocket audio hook, and stream the raw audio to Deepgram's live STT WebSocket for real-time transcription. Should demonstrate the full pipeline: Sinch call setup → audio stream extraction → Deepgram STT → transcript output.

Credentials likely needed

• SINCH_APPLICATION_KEY
• SINCH_APPLICATION_SECRET
• SINCH_PHONE_NUMBER
• DEEPGRAM_API_KEY

Reference

Sinch Voice API docs: https://developers.sinch.com/docs/voice/

---

Queued by PM from #142 on 2026-04-03

[github-actions]

Built in #166