xiaozhi-esp32-server
A backend server for the xiaozhi-esp32 open smart-hardware project, giving ESP32 devices speech recognition, model responses, text-to-speech and voice-fingerprint recognition over MQTT, UDP or WebSocket.
xiaozhi-esp32-server is a backend service for xiaozhi-esp32, an open smart-hardware project, letting a developer run their own server for ESP32 devices using the xiaozhi communication protocol. It is implemented in Python, Java and Vue, and communicates with devices over MQTT plus UDP, WebSocket, or an MCP access point.
It supports voiceprint recognition and a knowledge base, and offers two deployment modes: a minimal setup with configuration stored in files and no database, or a full setup with a database, multi-user and multi-agent management, and a control-panel interface.
A performance-testing tool measures the response speed of the configured speech-recognition, language, vision-language and text-to-speech modules, and the project documents its speech-recognition, model and speech-synthesis component options and their hardware requirements.