Voice_activated-RAG_chatbot

Voice_activated-RAG_chatbot is a long-term communication chatbot with persistent memory and document question-answering capabilities. Users can interact with the chatbot using either voice or text, and the system is bilingual—it understands both Hindi and English. The chatbot leverages Retrieval-Augmented Generation (RAG) to analyze user-uploaded documents along with past conversations, providing contextually relevant and coherent responses.

Features

• Bilingual Communication:
  Understands both Hindi and English spoken or typed input.

• Voice Activation:
  Users can interact via voice commands using built-in speech recognition and text-to-speech.

• Document Analysis:
  Upload documents (PDF, DOCX, or TXT) to have the chatbot extract and analyze their content.

• Persistent Memory:
  Past conversations are stored in a MySQL database. A vector database (Chromadb) is used for semantic search across stored conversations to provide relevant context.

• Retrieval-Augmented Generation (RAG):
  Uses an embedding model (e.g. mxbai-embed-large) and a local chat model (e.g. gemma2 via Ollama) to semantically search through and incorporate historical conversation data.

• Long-Term Communication:
  Maintains context over extended interactions for a natural, coherent dialogue experience.

Architecture & Technologies

• Streamlit:
  Provides the web-based user interface for chat, document upload, and controls.

• Ollama:
  Runs local AI models for both chat generation and embedding creation.

• MySQL:
  Stores conversation history persistently.

• Chromadb:
  An in-memory vector database used to build and query embeddings for past conversations.

• SpeechRecognition & Pyttsx3:
  Enable voice input and text-to-speech functionalities.

• Document Processing:
  Utilizes PyPDF2 for PDFs, docx2txt for DOCX files, and standard decoding for TXT files.

Installation

1. Clone the repository:

   git clone https://github.com/Manish-git-tech/Voice_activated-RAG_chatbot.git
   cd Voice_activated-RAG_chatbot

2. Set up a virtual environment (optional but recommended):

   python -m venv venv
   source venv/bin/activate      # For Linux/MacOS
   .\venv\Scripts\activate       # For Windows

3. Install dependencies:

   pip install -r requirements.txt

4. Set up MySQL:

   • Ensure you have a MySQL server running.
   • Create a database named AI_db3.
   • Create a table called conversations with at least the following columns:
     id (INT, primary key, auto-increment), prompt (TEXT), response (TEXT).

5. Configure Ollama:

   • Install and run Ollama locally.
   • Pull the required models:
     • For embeddings: ollama pull mxbai-embed-large
     • For chat: ollama pull gemma2
   • Verify that Ollama is running (typically on port 11434).

Usage

1. Run the Streamlit App:

   streamlit run GUI_chatBOT.py

2. Using the Chatbot:

   • Document Upload:
     Use the sidebar to upload a PDF, DOCX, or TXT file. The document’s text will be extracted and used as context for answering questions.

   • Voice or Text Input:
     Ask your question by typing it in the chat input or by using the voice buttons (for English or Hindi).

   • Persistent Memory:
     Past conversations are stored in the MySQL database and incorporated via Chromadb’s semantic search, so the chatbot can recall and use previous interactions to provide better answers.

Contributing

Contributions are welcome! If you have suggestions, bug fixes, or enhancements, please open an issue or submit a pull request.

Acknowledgements

• Thanks to the developers of Streamlit, Ollama, Chromadb, MySQL Connector, and other libraries used in this project.
• Credits to the GDSC club of IIT Indore for giving such a good project.