Urdu Virtual Assistant

This project is a comprehensive speech-to-speech model that understands and responds in Urdu. It integrates natural language processing with real-time information retrieval to provide a seamless conversational experience. The assistant can handle everyday conversations in Urdu and access relevant real-time information through the Tavily search engine.

Features

• Speech-to-Speech Interaction: Converts spoken Urdu queries into text, processes the input, and generates spoken Urdu responses.
• Real-Time Information Retrieval: Integrates with Tavily search engine to provide up-to-date information.
• Language Translation: Translates Urdu queries to English for processing and then translates responses back to Urdu.
• Text-to-Speech: Converts generated responses into speech using Google Text-to-Speech (gTTS).
• Customizable: Built with modular components, making it easy to extend or adapt to different use cases.

Technologies Used

• Streamlit: Frontend for the application.
• Google Text-to-Speech (gTTS): For converting text to speech.
• Hugging Face Transformers: Specifically, wav2vec2 for Urdu speech recognition.
• LangChain: For building conversational AI with Groq's Llama model.
• Tavily Search Engine: For retrieving real-time information.

Project Structure

├── app.py              # Main application code
├── requirements.txt    # Python dependencies
├── .env                # Environment variables
├── bolo_logo.png       # Logo image
├── audio.wav           # Temporary audio file storage
└── README.md           # Project documentation

Setup and Installation

1. Clone the Repository

git clone https://github.com/your-username/urdu-virtual-assistant.git
cd urdu-virtual-assistant

2. Install Dependencies

Make sure you have Python 3.8+ installed. Then, run:

pip install -r requirements.txt

3. Environment Variables

Create a .env file in the project root and add the following variables:

streamlit run app.py

The application should now be accessible at http://localhost:8501.

Usage

• Launching the Application: Once the app is running, you will see a title and logo at the top of the page.
• Recording a Query: Click the "Click to record" button to start recording your Urdu query. Click again to stop.
• Processing the Query: The app will convert the spoken query to text, process it, and then generate a spoken response.
• Listening to the Response: The AI's response will be played back in Urdu.

Acknowledgments

• Special thanks to the creators of Streamlit, LangChain, Hugging Face, and other open-source libraries used in this project.
• Thanks to Groq and Tavily for providing the necessary APIs and models.

Contributors

• Zohaib Saqib
• Ghulam Abbas
• Adeena Khudadad
• Abdur Rahman
• Ryyan Abdullah