[Suggestion] Multilingual voice agent with automatic Nova-3 language detection and per-language routing (Python)

[deepgram-robot]

What to build

A Python voice agent example that uses Deepgram Nova-3's automatic language detection to identify the caller's language and dynamically route the conversation — adjusting the LLM system prompt, TTS voice, and response language accordingly.

Why this matters

Multilingual voice agents are essential for global businesses and multilingual markets. Developers need a working example showing how to detect a caller's language from their first utterance, route to the appropriate language-specific configuration, and maintain a coherent multilingual conversation. Deepgram Nova-3 supports 36+ languages with automatic detection, making it the foundation for this pattern.

Suggested scope

• Python, using deepgram-sdk
• Deepgram Nova-3 with language=multi or detect_language=true
• Dynamic routing: detect language → set LLM system prompt in detected language → select appropriate TTS voice
• Support at least 3 languages (e.g., English, Spanish, Hindi)
• Deepgram Voice Agent API or direct STT + LLM + TTS pipeline

Acceptance criteria

• Runnable with minimal setup (clone, add API key, run)
• README explains the pattern clearly
• Uses current SDK version
• Demonstrates automatic language detection and routing for at least 3 languages
• Handles mid-conversation language switching

---

Raised by the DX intelligence system.