Server-side audio processing means that audio is processed on LiveKit servers or agents. This typically includes features like voice activity detection (VAD), speech-to-text, language model processing, and noise cancellation. These are applied after audio is sent from the client, and server-side VAD is enabled by default—you can fine-tune its settings to match your needs, and having noise cancellation on the server means you should disable it on the client to avoid unexpected issues. Client-side audio processing happens directly on the user’s device before audio is transmitted. This allows you to manually manage features like VAD (for example, implementing a push-to-talk interface where you control when the user is speaking). Client-side processing gives you more manual control, but generally, server-side is preferred for comprehensive and consistent processing, especially for AI-powered features. The main difference is where the audio processing occurs—on the user’s device before sending, or on the server after receiving. This affects control, quality, and which features are easiest to implement. Let me know if you’d like specific details or configuration guidance for your application. Sources: Voice Assistants | livekit/livekit_composite | DeepWiki | Turn detection | LiveKit Docs | AI noise cancellation | LiveKit Docs | Noise & echo cancellation | LiveKit Docs | Live Conversations with AI using ChatGPT and WebRTC