AgentX

AgentX is an advanced AI-powered assistant that integrates voice processing, document retrieval, conversational memory, and generative AI capabilities. This agent is designed to handle a wide range of tasks, from answering queries to processing voice commands.

---

Features

1. Voice Recognition

   • Converts speech to text using Vosk models.
   • Processes voice commands and generates audio responses via Google Text-to-Speech (gTTS).

2. Document Retrieval

   • Utilizes FAISS for efficient document embedding and retrieval.
   • Supports multiple document types for context-aware responses.

3. Generative AI

   • Powered by OpenAI GPT models for conversational capabilities.
   • Generates meaningful responses based on retrieved documents and context.

4. Conversational Memory

   • Maintains context across interactions using LangChain's ConversationBufferMemory.

5. Integration Ready

   • Modular design to integrate additional tools like Whisper API, advanced LangChain tools, or scheduling tasks.

---

Tech Stack

Tool	Purpose
LangChain	Conversational chains and memory handling
OpenAI GPT	Language generation and query answering
Vosk	Speech-to-text offline transcription
gTTS	Text-to-speech conversion
FAISS	Document embedding and vector retrieval
playsound	Audio playback
dotenv	Environment variable management
pytest	Unit testing framework
Logging	Structured logging for debugging

---

Project Structure

.
├── config
│   ├── constants.py        # Reusable constants
│   ├── settings.py         # Application settings and environment variables
├── core
│   ├── agent.py            # Main agent logic
│   ├── document_handler.py # Document loading and splitting
│   ├── memory.py           # Conversational memory handler
│   ├── voice_processor.py  # Voice processing logic
├── data
│   ├── documents           # Sample documents for testing
│   ├── audio               # Audio files (input/output)
├── model
│   └── vosk-model-small-en-us-0.15  # Speech recognition model
├── scripts
│   ├── run_agent.sh        # Script to run the agent
│   ├── setup_env.sh        # Script to set up the environment
├── tests
│   ├── test_agent.py       # Tests for agent.py
│   ├── test_memory.py      # Tests for memory handler
│   ├── test_voice_processor.py  # Tests for voice processing
├── Dockerfile              # Docker setup
├── README.md               # Project documentation
├── requirements.txt        # Python dependencies
└── main.py                 # Entry point for the application

---

Prerequisites

1. Python 3.11 or later
2. Vosk Model: Download and place the model in the model directory.
3. API Keys:
   • OpenAI API Key
   • Whisper API Key (Optional)

---

Installation

1. Clone the repository:

   git clone https://github.com/sattyamjjain/AgentX.git
   cd AgentX

2. Set up the environment:

   ./scripts/setup_env.sh

3. Create a .env file:

   OPENAI_API_KEY=<your_openai_api_key>
   WHISPER_API_KEY=<your_whisper_api_key>
   DEBUG=True
   

4. Run the application:

   python3 main.py

---

Testing

Run the test suite using pytest:

pytest tests/

---

Usage

1. Running the Agent:

   python3 main.py

2. Interactive Voice Commands:

   • Speak into the microphone, and AgentX will respond.

3. Document Retrieval:

   • Place .txt files in data/documents to make them available for querying.

---

Future Enhancements

1. Integration with Whisper API for high-quality transcription.
2. Multi-modal capabilities for image and video processing.
3. Improved conversational flows with dynamic memory management.

---

Contributors

• Sattyam Jain
• Open to contributions!

---

Support

For issues or feature requests, please open an issue on the GitHub repository.