[Example] 051 — Next.js Streaming STT + TTS with Deepgram via Vercel AI SDK

[github-actions]

New example: Next.js Streaming STT + TTS with Deepgram via Vercel AI SDK

Integration: Vercel AI SDK + Next.js | Language: TypeScript | Products: STT, TTS

What this shows

A full-stack Next.js 15 App Router application that captures microphone audio in the browser and streams it to Deepgram for real-time transcription (nova-3), with live interim results displayed as the user speaks. Includes text-to-speech playback using Deepgram Aura 2 via the Vercel AI SDK's provider-agnostic generateSpeech() function. Demonstrates secure temporary API key provisioning so the main key never reaches the browser.

Required secrets

None — only DEEPGRAM_API_KEY required

Closes #24

---

Built by Engineer on 2026-04-01

[github-actions]

Code Review

Overall: APPROVED

Tests ran ✅

Test 1: Transcription via Vercel AI SDK...
  OK  transcription (321 chars, found: spacewalk)
Test 2: Text-to-Speech via Vercel AI SDK...
  OK  TTS (251564 bytes linear16)
Test 3: Deepgram SDK project access (for temp key creation)...
  OK  found 1 project(s)

All tests passed

Integration genuineness

Pass — The Vercel AI SDK integration is genuine:

• @ai-sdk/deepgram provider imported and used for both generateSpeech() (TTS) and transcribe() (STT)
• @deepgram/sdk used for manage.getProjects() and keys.createKey() to mint temporary browser keys
• Browser opens a real WebSocket to wss://api.deepgram.com/v1/listen with nova-3
• Tests make real API calls (exit 2 on missing credentials, not fake passes)
• .env.example lists DEEPGRAM_API_KEY — the only credential needed since the Vercel AI SDK uses it through the provider

Code quality

• Official Deepgram SDK used throughout — no raw HTTP calls to Deepgram
• No hardcoded credentials; temporary browser keys have a 10-second TTL with usage:write scope only
• Error handling covers: missing API key (both routes), WebSocket errors, TTS failures, empty transcripts
• Credential check in tests/test.js runs before any SDK imports (lines 11–21)
• Both API routes validate DEEPGRAM_API_KEY before proceeding
• Clean float32↔int16 PCM conversion for both mic capture and TTS playback

Documentation

• README covers: what you'll build, prerequisites, env vars with console links, install/run instructions, key parameters table, architecture diagram
• .env.example present and complete

---

✓ All checks pass. Ready for merge.

---

Review by Lead on 2026-04-01