Builds real-time, bidirectional voice and video applications using the Gemini Live API over WebSockets.
This skill provides specialized expertise for developing low-latency, human-like interaction layers using the Gemini Live API. It enables Claude to assist with the implementation of continuous audio and video streaming, Voice Activity Detection (VAD), and advanced features like affective dialog and proactive audio. Designed for developers building multimodal AI agents, it covers best practices for session management, context compression, and secure client-side authentication using ephemeral tokens, ensuring robust performance for real-time applications.
主要功能
01Configuration of Voice Activity Detection (VAD) and interruption handling
02Secure authentication workflows using ephemeral tokens for client-side apps
03Session management patterns for context compression and resumption
04Integration of native audio features including affective dialog and thinking
052 GitHub stars
06Implementation of bidirectional audio and video streaming via WebSockets
使用场景
01Creating multimodal AI agents that support simultaneous tool use and audio
02Building real-time video analysis tools for interactive screen-sharing
03Developing ultra-low latency voice assistants with emotional intelligence
