Kevinobote/Kiswahili-Kitukuzwe

# Kiswahili Audio Processing Pipeline A Novel Kiswahili Audio Processing Pipeline: An Integrated Approach to Speech Recognition, Sentiment Analysis, and Text Summarization. > **Note**: This is a research project developed as part of a dissertation on Kiswahili NLP processing. ## 🎯 Overview This project implements a complete audio processing pipeline that: - **Speech Recognition**: Converts Kiswahili audio to text using Wav2Vec2 - **Sentiment Analysis**: Analyzes emotional tone using DistilBERT - **Text Summarization**: Generates concise summaries using T5 ## 🏗️ Architecture ``` ├── main.py # FastAPI application entry point ├── models/ │ ├── pipeline_manager.py # Core ML pipeline orchestration │ └── __init__.py ├── schemas/ │ ├── response_models.py # API response models │ └── __init__.py ├── frontend/ │ ├── index.html # Main web interface │ └── static/ │ ├── app.js # Frontend JavaScript │ └── style.css # Styling ├── requirements.txt # Python dependencies ├── setup_env.sh # Environment setup script └── README.md ``` ## 🚀 Quick Start ### 1. Setup Environment ```bash # Make setup script executable and run chmod +x setup_env.sh ./setup_env.sh ``` ### 2. Activate Environment ```bash source sema-deployed/bin/activate ``` ### 3. Start Application ```bash python main.py ``` ### 4. Access Application Open your browser and navigate to: `localhost` ## 🔧 Manual Setup If you prefer manual setup: ```bash # Create virtual environment python3 -m venv sema-deployed source sema-deployed/bin/activate # Install dependencies pip install -r requirements.txt # Run application python main.py ``` ## 📱 Features ### Web Interface - **Audio Upload**: Support for various audio formats - **Live Recording**: Record audio directly in browser - **Real-time Processing**: Asynchronous pipeline processing - **Responsive Design**: Works on desktop and mobile devices ### API Endpoints - `POST / …