Real-time speech detection with dynamic chunking and AI suggestions
This demo uses energy-based VAD to detect speech segments dynamically. Unlike fixed-time chunking, VAD finds natural speech boundaries and sends complete utterances to AI for better context and suggestions.