AGi-Test

A testbed for the Amazing GRACE Infrastructure, enabling rapid prototyping and validation of advanced AI agent capabilities. Part of the OppaAI ecosystem.

🎯 Overview

AGi-Test is a sandbox environment for developing and testing cutting-edge AI agent features, with a focus on:

• Voice-Enabled AI - Japanese-speaking companions with neural TTS and voice blending
• Function-Calling Agents - LLM-powered chatbots with external tool integration
• Multimodal AI - Computer vision and real-time visual understanding
• Local LLM Inference - Privacy-first AI using Ollama and open-source models

✨ Features

🎤 Voice Chatbot (kokoro_chatbot.py / coqui_chatbot.py)

Interactive voice-based AI companion with advanced speech synthesis:

• LLM Backend: Ollama with Gemma3-Code-Reasoning-4B
• TTS Engine: Kokoro or Coqui with neural voice blending
• Persona: Customizable Japanese-speaking companion
• Voice Profiles: Neutral and ASMR voice blending using PyTorch
• Features:
  • Real-time streaming responses
  • Session-based chat history
  • Japanese-only output mode
  • Multi-voice synthesis and blending
  • Interactive chat loop (type "exit" to quit)

Requirements:

• Ollama with Gemma3-Code-Reasoning-4B model
• Kokoro/Coqui TTS models in /models
• CUDA-compatible GPU (recommended for voice blending)

🔧 Function-Calling Chatbot (archive/agi_v202.py)

Async chatbot with integrated external tools:

• Tools: DuckDuckGo search, weather lookup, aurora forecasts, date/time
• Architecture: Async chat loop with tool-call handling
• Memory: Persistent chat history per session

👁️ Multimodal AI (archive/agi_v203.py)

Computer vision integration for real-time scene understanding:

• Vision Model: Moondream2
• Input: Webcam or image files
• Capabilities: Object detection, scene analysis, visual reasoning
• Interaction: Combined text and image-based queries

🚀 Quick Start

Prerequisites

• Python 3.10 or higher
• Ollama installed and running
• NVIDIA GPU with CUDA support (optional but recommended)

Installation

# Clone the repository
git clone https://github.com/OppaAI/AGi-Test.git
cd AGi-Test

# Install dependencies
pip install -r requirements.txt

# Or use uv for faster installation
pip install uv
uv sync

Running the Chatbots

# Voice chatbot with Kokoro TTS
python kokoro_chatbot.py

# Voice chatbot with Coqui TTS
python coqui_chatbot.py

# Function-calling chatbot (archived)
python archive/agi_v202.py

# Multimodal AI with webcam (archived)
python archive/agi_v203.py

📁 Project Structure

AGi-Test/
├── kokoro_chatbot.py          # Main voice chatbot (Kokoro TTS)
├── coqui_chatbot.py           # Alternative voice chatbot (Coqui TTS)
├── requirements.txt           # Python dependencies
├── models/                    # Pre-trained voice models
│   ├── neutral_voice.pt
│   ├── ASMR_voice.pt
│   └── [other voice models]
├── assets/                    # Sample audio and metadata
│   ├── sample.wav
│   ├── bender.wav
│   └── *.json
└── archive/                   # Previous implementations
    ├── agi_v202.py           # Function-calling chatbot
    ├── agi_v203.py           # Multimodal AI
    └── eye.py                # Vision system

🔧 Key Dependencies

Package	Purpose
ollama	Local LLM inference
kokoro / coqui-tts	Neural text-to-speech
torch	Deep learning framework
transformers	Pre-trained models
opencv-python	Computer vision
realtimetts	Real-time audio streaming
duckduckgo-search	Web search integration

See requirements.txt for the complete dependency list.

🎓 Use Cases

• AI Companion Development - Build and test voice-based AI assistants
• Voice Synthesis Research - Experiment with neural TTS and voice blending
• Agent Prototyping - Rapidly develop and validate new agent capabilities
• Multimodal AI - Combine language and vision for richer interactions
• Privacy-First AI - Run everything locally without cloud dependencies

🔮 Future Roadmap

• Comprehensive error handling and logging
• Database storage for conversation history
• Dynamic model selection and switching
• Flexible persona management system
• Extended tool ecosystem
• Web UI for easier interaction
• Model fine-tuning capabilities

📚 Related Projects

• OppaAI/AGi — Main AGi project
• OppaAI/MCP-Client — MCP Client implementation

⚖️ License & Legal

This project is licensed under the GNU General Public License v3.0. See LICENSE for details.

⚠️ Important Notice

The voice cloning and synthesis features are strictly for personal, educational, and hobby purposes only. Any use for inappropriate, criminal, or unauthorized commercial activities is strictly prohibited.

🙏 Acknowledgments

Special thanks to:

• RealtimeTTS - Real-time audio streaming library
• Kokoro - Neural TTS framework
• Coqui TTS - Open-source text-to-speech
• Ollama - Local LLM inference
• Moondream - Vision model
• DuckDuckGo Search - Web search API
• OpenCV - Computer vision library
• Transformers - Pre-trained models
• The open-source AI community

---

Questions or contributions? Visit the OppaAI GitHub or open an issue!