Xiaozhi ESP32 Server is a Java enterprise-grade server for the Xiaozhi AI chatbot hardware (xiaozhi-esp32), provided by the xiaozhi-esp32-server-java project. Built with Spring Boot 3 + Vue 3, it delivers an integrated "management console + server" solution.

Features

Device monitoring: online status, live conversations, and one-stop device binding management;
Voice customization: multiple built-in ASR/TTS engines (local VOSK / Sherpa + Edge TTS / Volcano Engine, etc.) with voice and speaking-rate adjustment;
Persona switching: custom agent personas and prompt templates with one-click role switching;
Conversation history: full dialogue history with audio replay, plus knowledge-base augmented Q&A;
Multi-client access: ESP32 devices (WebSocket), mini program, and web console.

Prerequisites

Before installing this app, install a MySQL (8.x) or MariaDB (10.6+) app from the app store first. Select the installed database in the "Database Service" field during installation — 1Panel creates the database and user automatically, and the server applies its schema via Flyway migrations on first start.

After Installation

Wait until all containers become healthy (the first start runs database migrations, about 1–2 minutes);
Open the console at http://SERVER_IP:console_port and sign in with the default account admin / password 123456 — change the password immediately after the first login;
Configure an LLM API key under "Model Management" in the console to start chatting (no keys are bundled);
ESP32 devices: point the device firmware OTA URL to http://SERVER_IP:api_port/api/device/ota, and the device will establish the real-time conversation over the WebSocket port.

Tip: speech recognition uses the bundled local model and speech synthesis uses the free Edge TTS out of the box; switch to third-party APIs in the console for higher quality.

Ports

Port	Purpose
Console port (default 8084)	Web management console
API / OTA port (default 8091)	Server API and device OTA
Device WebSocket port (default 8092)	Real-time voice conversation for ESP32 devices

Resource Requirements

Recommended: 2 CPU cores and 4 GB+ available memory (the server and dialogue services reserve 2 GB each).

Data & Upgrades

Business data lives in the selected MySQL/MariaDB app — back it up together with that database app;
Conversation audio (data/audio), uploads (data/uploads), Redis cache and logs are kept under the version directory's data/ folder.

README_en.md Unescape Escape

What is Xiaozhi ESP32 Server?