@ 非常重要的事情！ The most important thing!

Caution

Due to financial challenges and uncertainties regarding the project's future direction, I have made the difficult decision to discontinue further development. This project was initially conceived on August 12, 2024, officially launched on January 18, 2025, and is now being suspended as of October 30, 2025. I would like to extend my deepest gratitude to all supporters, contributors, and developers who have participated in secondary development throughout this journey.

由于资金方面的挑战以及对项目未来发展方向的不确定性，经过慎重考虑，我遗憾地决定暂停本项目的更新。该项目始于2024年8月12日，正式立项于2025年1月18日，今日（2025年10月30日）正式停止更新。在此，我向一直以来支持本项目、参与二次开发及做出贡献的各位致以最诚挚的感谢。

---

🤖 AI Desktop Pet

English | 简体中文
📚 API Documentation

---

Caution

The program's kernel is updating...

The in-program kernel update package relates to the programming language for plugins and extensions. The programming language will be replaced with our self-developed one (of course, Python and other languages are still supported). This new programming language is designed for regular users.

Supported for Chinese Programming

---

---

🌟 Project Overview

This is a cross-platform desktop pet driven by artificial intelligence, supporting highly customizable appearance and interactive behaviors. The project employs a modular design, integrating the following core features:

• 🎭 Multi-form Character Support (2D Models)
• 🗣️ Intelligent Voice Interaction (Speech Recognition + NLP)
• ✨ Physics-Engine-Driven Realistic Behavior Simulation
• 🔌 Plugin-Based Extension System (Python)

---

🚀 Key Features

🌈 Interactive Features

graph TD
    A[User Input] --> B{Input Type}
    B -->|Speech| C[Whisper Speech Recognition]
    B -->|Touch| D[Contact Point Analysis]
    C --> E[Natural Language Processing]
    D --> F[Physical Behavior Response]
    E --> G[Multi-modal Response Generation]
    G --> H[TTS Speech Synthesis]
    H --> I[Character Animation]
    F --> I

Loading

---

🛠️ Quick Deployment

System Requirements

Recommended System Requirements

• Windows 10/11 x64
• NVIDIA GPU (RTX 20 series or higher recommended)

Least Requirements

• Windows 10/11 x32
• i5-4 Series

Installation Steps

1. Download the latest version from the Releases page
2. Extract to the target directory (recommended to use an English path)

   Ai Desktop Pet.exe

3. API Key Configuration (Guidelines)

---

🔑 API Key Configuration

Alibaba Cloud Bailian Large Model

1. Log in to the Bailian Console
2. Create an application → Obtain API Key
3. Fill in the configuration → AI → Cloud Inference

Xunfei Speech Service

1. Log in to the Xunfei Cloud Console
2. Create a Streaming Speech Recognition application → Obtain API Information
3. Fill in the configuration → AI → Cloud Inference

---

🧩 Plugin Development

Official Plugin Market

Plugin Name	Description	Version
Plugin Manager	Manage installed plugins	0.0.2
Raising System	Intelligent Training and Growth System	0.5.0

---

🔧 Local Deployment Guide

🚨 Important Note: Local deployment requires 1.42GB for BASIC RUNTIME

⚠️ The least space requirement is 3.22GB for FULL

---

🌐 Service Architecture Topology

graph LR
    A[Main Program] --> B[Whisper Speech Recognition]
    A --> C[Ollama Language Model]
    A --> D[GPT-SoVITS Speech Synthesis]
    B --> E[Local CPU Inference]
    C --> F[Local GPU Inference]
    D --> G[Local GPU Inference]

Loading

---

🎙️ Speech Recognition Deployment (Whisper)

📦 Environment Setup

# Enter the driver directory
cd AA-package-driver

# Install dependencies (recommended to use a virtual environment)
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt

🚀 Start the Service

python Whisper_api.py

---

🗣️ Speech Synthesis Deployment (GPT-SoVITS)

📂 Model Directory Structure

gsv/
├── Chocolate;ja/
│   ├── ご主人様のお父様にいつかうまいって言わせてみせるって.wav
│   ├── Chocolate-e60.ckpt
│   └── Chocolate_e10_s3600.pth
└── Maple;ja/
    ├── ところで、花椒。パンプキンケーキに合わせて茶葉を選んでみたけど。.wav
    ├── Maple-e100.ckpt
    └── Maple_e10_s4510.pth

⚠️ Key Requirements

1. Folder naming format: Character Name;Language Code (e.g., Chocolate;ja)
2. WAV file names must match the complete text content of the corresponding speech
3. Must include the following three file types:
   • .wav reference audio
   • .pth generation model
   • .ckpt fine-tuning weights

---

🤖 Local Large Model Deployment (Ollama)

🛠️ Installation and Configuration

1. Download the Ollama Windows version
2. Set environment variables (optional):

   # Modify the model storage path
   [Environment]::SetEnvironmentVariable("OLLAMA_MODELS", "<Your Model Path>", "User")

3. Restart the terminal to apply the configuration

🧠 Recommended Model Configurations

Warning

All the command must run on Windows PowerShell platform

VRAM Capacity	Recommended Model	Startup Command
1-4GB	Qwen2.5-0.5B	ollama run qwen2.5:0.5b
4-6GB	Qwen2.5-1.5B	ollama run qwen2.5:1.5b
6-9GB	Qwen2.5-3B	ollama run qwen2.5:3b
9-15GB	Qwen2.5-7B	ollama run qwen2.5:7b
18-22GB	Qwen2.5-14B	ollama run qwen2.5:14b
22-26GB	Qwen2.5-32B	ollama run qwen2.5:32b

---

🔄 API Interface Configuration

📡 Server Endpoint Settings

{
  "api": "heavynotfat",
  "model": "qwen2.5:3b",
  "messages": "{{messages}}",
  "tools": "{{tools}}",
  "Answer Index": "message.content"
}

API Configuration Screenshot

🔑 Key Field Descriptions

Field	Required	Description
messages	✓	Chat history (automatically filled by the program)
Answer Index	✓	Response parsing path (e.g., message.content)
Model	✓	Model name
API-Key	x	API key
tools	x	Available tools list (in JSON format)

---

📜 Open Source License

This project uses the GPL-3.0 License, with the following key restrictions:

• Modified code must be open-sourced
• Derivative works must explicitly credit the original copyright
• Cannot be used for military purposes

For commercial use, contact the author to obtain a commercial license.

---

📧 Contact Developer: 2953911716@qq.com

🌐 Official Community: https://github.com/HeavyNotFat/Agentic-AI-Desktop-Pet/discussions