Hey everyone! Can anyone share their STT -> LLM -> TTS stack for real-time voice AI & RAG use case optimized for latency? Any tips welcome! 🙏🏼