We built a voice AI agent in Rust. EchoKit server is an open-source framework that connects speech input, LLM reasoning, and speech output — so you can build local or hybrid voice assistants with real-time interaction. It's super easy to set up. Current features: 1. Supports both traditional ASR → LLM → TTS pipelines and end-to-end models like Gemini Live. 2. Voice Activity Detection (VAD) for low-latency, streaming voice input. 3. MCP support for external tools. 4. Works with both ESP32 hardware and web clients. 5. Streaming TTS output for GPT-Sovits Would love to hear what kind of projects people would build with this or any feedback. |