xinnan-tech

xiaozhi-esp32-server: Run a Voice-AI Backend for ESP32 Devices

A self-hostable backend for xiaozhi-esp32 voice-enabled hardware, with speech, language-model, device-control, and management features. It suits builders who want to configure their own ESP32 service and connect AI providers, but the project warns against production use.

ActiveJavaScriptMITAISelf HostedMCP Server
View on GitHub Compare
Stars
10.7k
Forks
3.7k
Stars, last 30 days
+429
Last commit
1 day ago
Contributors
100+

Our take

Activeregular commits and releases

Health score

75/100

How it is scored
Activity
76
Community
87
Issues
81
Pull requests
55

+429 stars in the last 30 days

  • More contributors than 78% of the projects we track
  • Merges more pull requests than 76% of the projects we track

This is a feature-rich, actively maintained backend for the xiaozhi-esp32 ecosystem, but its own warning that it is incomplete and has not passed a security assessment makes it unsuitable for production deployment as presented.

Good fit if

  • You already use compatible xiaozhi-esp32 hardware and want to run or customize its backend.
  • You need choice among speech, LLM, memory, and tool integrations, and can manage their configuration.
  • You are evaluating or developing the system in a protected environment, where its MIT license and recent project activity are useful advantages.

Look elsewhere if

  • You need a backend the project itself endorses for production or public-internet deployment without additional security work.
  • You need a general-purpose voice assistant backend that is not tied to the xiaozhi-esp32 device ecosystem.
  • You want a single turnkey setup without configuring model and speech services or choosing between the documented deployment modes.
All health signals
Last commit2026-10-07 (1 day ago)
Commits, last 90 days100+
Releases, last 12 months13 (latest v0.9.7, 2026-09-29)
Contributors100+ (top contributor: 34% of commits)
Issues closed, last 90 days28+ (typically closed in 1 day)
Pull requests merged, last 90 days55 (typically merged in 1 day)
Project age20 months

Checked on 2026-10-08 with the GitHub API.

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

11,294

Forks

3,843

Change since Nov 14, 2026: +7 stars·+2 forks

Overview

xiaozhi-esp32-server provides the backend for xiaozhi-esp32, an open-source ESP32 hardware project. It handles voice and AI interactions between devices and configurable services, so builders can operate their own device-control server instead of relying on a hosted backend.

The project supports MQTT+UDP and WebSocket communication, alongside an optional web control console. It is intended for people with compatible ESP32 hardware who are comfortable configuring speech, model, and other service integrations.

Key Features

  • Connects ESP32 devices over MQTT+UDP or WebSocket, with HTTP endpoints also listed in the project architecture.
  • Integrates multiple speech-recognition, speech-synthesis, and language-model providers, including local options for some components.
  • Supports visual-model interactions, voice activity detection, and speaker voiceprint recognition.
  • Offers device and tool actions through IoT commands, MCP integrations, plugins, and custom functions.
  • Provides optional knowledge retrieval through RAGFlow and memory integrations.
  • Includes a web console for users, system settings, and device management in the full-module installation.
  • Documents Docker and source-code deployment paths, with a simpler server-only setup as an option.

Use Cases

  • ESP32 device owners who want to replace a hosted backend with a self-managed service.
  • Makers building voice-controlled home devices that need model-driven conversations and IoT command dispatch.
  • Developers prototyping multilingual or multimodal voice assistants with configurable ASR, LLM, and TTS providers.
  • Teams exploring MCP-connected tools or knowledge-backed responses on an ESP32-based assistant.

What you need

Detected in the repository

  • JavaScript (main language on GitHub)
  • Automated checks on GitHub Actions

License in plain words

MITpermissive

  • Commercial use: yes
  • Modify and redistribute: yes
  • You must keep: the copyright and license notice
  • Share your changes: no

A summary, not legal advice: the LICENSE file is what applies.

Getting Started

Choose either the simplified deployment or the full-module deployment. The repository documents Docker and source-code options for both; see the README and FAQ for setup and configuration details.

Considerations

  • The project explicitly says functionality is not complete and that it has not passed a network security assessment. It advises against production use and calls for protection when deployed on a public network.
  • Speech recognition, model responses, and speech synthesis may depend on third-party APIs. The project disclaims commercial relationships with those providers and responsibility for their service quality or payment safety.
  • The full installation uses a database, while the simplified option stores data in configuration files. The README gives different resource guidance depending on deployment mode and whether local FunASR is used.
  • The README describes development as ongoing. Although the repository is active, with regular recent commits and releases, adoption still involves a changing feature set.

Found this useful?

Share it with someone who would like xiaozhi-esp32-server.

Comparisons