This project provides a modular, flexible, and elegant Chinese voice dialogue robot and smart speaker system, targeting makers and hackers in China. It integrates advanced AI capabilities like ChatGPT for multi-turn conversations and offers unique features such as brain-computer interface (BCI) wake-up, aiming to enable personalized smart speaker experiences.

How It Works

The system operates on a pipeline: wake-up word detection (offline or BCI), Automatic Speech Recognition (ASR) to convert speech to text, Natural Language Understanding (NLU) for parsing, skill matching to identify the appropriate plugin, plugin execution, Text-to-Speech (TTS) synthesis, and finally, audio playback. This modular design allows each component, including ASR, TTS, and dialogue engines, to be independently customized or extended, supporting various third-party integrations and online/offline dialogue models.

Quick Start & Requirements

• Install/Run: python3 wukong.py
• Prerequisites: Python >= 3.7 and < 3.10.
• Supported Platforms: Intel Chip Macs (not M1), 64bit Ubuntu, Raspberry Pi series, Pine 64, Intel Edison, and Windows with WSL.
• Setup: Configuration involves creating a user-specific config file.
• Demo: A backend management demo is available at https://bot.hahack.com (user: wukong, pass: wukong@2019).

Highlighted Details

• Supports multiple Chinese ASR and TTS engines (Baidu, iFlytek, OpenAI Whisper, VITS, etc.).
• Integrates local (AnyQ) and online (ChatGPT, Turing) dialogue bots.
• Offers offline wake-up (Porcupine, Snowboy) and novel wake-up methods (Muse BCI, shake-to-wake).
• Enables smart home integration with platforms like HomeAssistant, XiaoAi, and Siri.
• Includes a backend management system for remote control and monitoring.

Maintenance & Community

• The project has seen significant adoption, with over 13,000 installations and 700,000 wake-ups as of March 2023.
• Community support is available via QQ channels and groups.
• The primary developer is 潘伟洲.

Licensing & Compatibility

• The project is released under a permissive license, suitable for commercial use and closed-source linking.
• The README states no affiliation with Tencent Dingdang or Ubtech Wukong projects.

Limitations & Caveats

The project explicitly states it is for personal learning and research, with no liability for any losses incurred from its use. It also notes that Intel Macs are supported, but M1 Macs are not. Users are advised to use their own API keys for services due to usage limits.