LiveKit handles voice input through a detailed audio pipeline that includes: • Voice Activity Detection (VAD): This helps spot when someone is actually speaking in the audio stream. • Speech-to-Text (STT): It turns what you say into text using STT models. • Text-to-Speech (TTS): Converts responses back into audio, so you can hear them, using TTS models. • Real-time Capabilities: Thanks to WebRTC, audio transfers happen super quickly with minimal delay. • Customization: Developers can pick their favorite STT providers, tweak TTS settings, and connect with different LLMs to fit their needs. Depending on the setup, audio formats like PCM, ALAW, and ULAW are used. The process involves streaming audio continuously, using VAD to efficiently detect speech, and ensuring low-latency processing for smooth, natural interactions. If you'd like, I’m here to help guide you through implementing any part of these features—just let me know! Sources: Agents Core Architecture | livekit/livekit_composite | DeepWiki | Voice Assistants | livekit/livekit_composite | DeepWiki | Voice Assistants | livekit/livekit_composite | DeepWiki | Agent speech | LiveKit Docs