MOMOKA is an AI Agent-type Discord Bot that acts as your intelligent Discord companion! 🤖✨ Unlike traditional bots that simply respond to commands, MOMOKA operates as an autonomous AI agent capable of making decisions, understanding context, and proactively assisting your community. It combines the power of advanced AI chat with autonomous capabilities, seamless music playback, local image generation, and essential utility commands—all in one sleek package. Perfect for communities that want an intelligent, self-aware bot that can think and act independently!

Built-ins:

• Built-in image generation engine - Fully integrated diffusers-based image generation pipeline (see MOMOKA/generator/image). No external services required!
• Integrated Style-Bert-VITS2 TTS engine - The complete Style-Bert-VITS2 source code is integrated into this project. No external API server needed! See MOMOKA/generator/tts and NOTICE for details.

• 🤖 AI Chat (LLM) - Simply mention the bot with @<bot name> to start chatting! Supports multiple AI models including OpenAI GPT-4, Google Gemini, NVIDIA NIM, and local KoboldCPP with automatic API key rotation to handle rate limits seamlessly
• 🎵 Music Playback - Play music from YouTube, Spotify, and more in voice channels
• 🎨 Image Generation (Built-in) - Fully integrated diffusers-based image generation engine. No external services required! Place models at models/image-models/<image model名>/<image model名>.safetensors (optional VAE/LoRA and model.json).
• 🗣️ Text-to-Speech (Built-in) - Fully integrated Style-Bert-VITS2 engine - The complete Style-Bert-VITS2 source code is built into this project. No external API server needed! Place models at models/tts-models/<tts model名>/<tts model名>.safetensors (or G_*.pth with matching config.json). Optional pyopenjtalk dictionary and style_vectors.npy are supported. See NOTICE for integration details.
• 📊 Game Tracking - Track stats for Rainbow Six Siege and VALORANT
• 🔔 Notifications - Get notified about earthquakes and Twitch streams
• 🎲 Utilities - Dice rolls, timers, media downloads, and more!

1. Clone the repository

   git clone https://github.com/coffin399/ProjectMOMOKA.git
   cd ProjectMOMOKA

2. Install dependencies

   pip install -r requirements.txt

3. Configure the bot

   • Copy config.default.yaml to config.yaml
   • Fill in your bot token and API keys

• (Optional) Place local models:

  Directory Structure:

  models/
  ├── image-models/
  │   └── <image model名>/
  │       └── <image model名>.safetensors
  │       └── (optional) VAE, LoRA, model.json
  └── tts-models/
      └── <tts model名>/
          └── <tts model名>.safetensors
          └── (optional) config.json, style_vectors.npy
  

  Examples:

  • Image models: models/image-models/my-model/my-model.safetensors
    • Optional: VAE, LoRA, and model.json in the same directory
  • TTS models: models/tts-models/my-voice/my-voice.safetensors
    • Alternative: G_*.pth with matching config.json
    • Optional: pyopenjtalk dictionary and style_vectors.npy

• Configure options in config.yaml (e.g., default image model, TTS defaults)

4. Run the bot

   Windows (Recommended): Use the all-in-one batch file that handles virtual environment setup and package installation automatically:

   startMOMOKA.bat

   Manual start (Linux/Mac or if you prefer):

   python main.py

For detailed documentation, please check the language-specific README files:

• 🇯🇵 日本語詳細ドキュメント (Japanese Detailed Documentation)
• 🇺🇸 English Detailed Documentation

Key guides inside the docs include:

• Configuring built-in image generation and model management
• Tips for Stable Diffusion prompt crafting

This project integrates source code from the following open-source projects:

• Style-Bert-VITS2 - Text-to-speech engine integrated into MOMOKA/generator/tts. The Style-Bert-VITS2 source code is built into this project under AGPL-3.0 and LGPL-3.0 licenses. See NOTICE and MOMOKA/generator/tts/LICENSE_SBVITS2* for details.