explain me the code flow how the events received from stt are being passed to llm and then to the tts?

Thanks for reaching out, Aditya. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

its general

Hey there! Let me break down how LiveKit processes events from speech to text, through AI, and back to speech in a super friendly way. • STT Node: Your audio gets transformed into text using the

stt_node()

function - pretty cool, right? • LLM Node: Then, our smart AI takes that text and thinks up a clever response with the

llm_node()

function. • TTS Node: Finally, the AI's response gets turned back into speech using the

tts_node()

function. Here's how it all comes together behind the scenes: 1. The system shifts into "thinking mode" - just like when you're pondering a great idea! 2. It runs the AI inference by calling

perform_llm_inference()

with the agent's

llm_node

. 3. The AI output gets split into two streams using

tee

. 4. If audio is enabled, it converts the text to speech by calling

perform_tts_inference()

with the agent's

tts_node

. 5. The system waits for a green light before sharing the response. The best part? This setup is super flexible, so developers can tweak things at any point in the process. How awesome is that? Sources: Voice pipeline nodes | LiveKit Docs | GitHub::livekit::agents | livekit-agents/livekit/agents/voice/agent_activit...