Thanks to visit codestin.com
Credit goes to github.com

Skip to content

kleinlee/MatesX

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

35 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MatesX

Memory · Expression · Motion · Multi-platform · Lightweight

English | 中文版

🎯 Project Mission

  • Enable individual users to customize their own AI companions
  • Support ultra-high concurrency digital human services targeting massive C-end users

Three Core Objectives:

  1. ✅ Large-scale C-end Digital Human Conversation Management

  2. ✅ Memory, emotion, Expression & Motion Management

  3. ✅ Cross-platform Support for Desktop, APP, Mini Programs with Ultra-lightweight Architecture


🚀 Core Features

1. 🗣️ End-to-End Digital Human Conversation Engine

VAD (Voice Activity Detection) → ASR (Automatic Speech Recognition) → LLM (Large Language Model) → TTS (Text-to-Speech) → Digital Human Animation

2. 🧠 mem0-Level Memory Functionality

  • Fusion of long-term and short-term memory, supporting personalized dialogue history storage and contextual understanding
  • Dynamic user profiling, persistent memory with cross-device synchronization

3. 😊 Real-time Emotion Analysis

  • Multimodal emotion recognition based on voice tone, semantic content, and dialogue rhythm
  • Outputs emotion tags (happy/sad/angry/surprised, etc.) to drive facial expressions and vocal intonation

4. 🎭 Free Expression & Free Motion

  • Driven by large-model algorithms, supports skeletal/BlendShape animation
  • Automatically matches facial expressions and body language in real-time based on dialogue content and emotional state

📱 Multi-platform Support

Platform Status Tech Stack Preview
Windows electron exe link
macOS electron
Android app webview apk link
WeChat Mini Program webview 小程序:MatesX数字生命
Web WebGL rendering support web link

💡 All platforms share the same core engine, code reuse rate > 99%, extremely lightweight (core module < 5MB)


🛠️ Quick Start

git clone https://github.com/kleinlee/MatesX.git
cd MatesX
pip install -r requirements.txt

Deploy cloud-based conversation & speech models (Alibaba DashScope)

Modify utils/dashscope.py:

# Configure LLM-specific API
DASHSCOPE_API_KEY = ""     # Obtain API-key from Alibaba Cloud Bailian
DASHSCOPE_TOKEN_URL = "https://dashscope.aliyuncs.com/api/v1/tokens"
DASHSCOPE_LLM_URL = "https://dashscope.aliyuncs.com/compatible-mode/v1"
# Configure your domain server (or use localhost for local only)
HOST_URL = "http://localhost:8000"
# Configure MatesX Avatar key
MatesX_key = ""           # Obtain key from matesx.cn
# Optional: config if using Tencent cloud TTS
USE_TENCENT_TTS = True  # default: False (prefer Alibaba Cloud voice by default)
class TencentTtsConfig:
    SECRET_ID = ""
    SECRET_KEY = ""
    APP_ID = 

Start service:

python main.py

Then open http://localhost:8000/web/home.html to enjoy!

Custom Avatars

This project does not offer underlying algorithm implementations but focuses on lightweight yet comprehensive applications.

For individual enthusiasts, get free credits on Matesx official web application.

For developers, receive 3 free credits by contacting us. please refer to the API for API calling. Use matesx.cn to manage your avatars.

Compilation

support Windows, macOS, Android, Mini-Program

Refer to the platform folder

High Concurrency & Cloud Sync

  • Upgrade to cloud database
  • Host resources on OSS

License

Apache License 2.0

Contact

join WeChat group Join QQ group
MatesX 官方微信 MatesX 官方QQ