sayleee1/speech-to-text-pipeline

Web app to collect Amharic speech data for NLP model training # 🗣️ Speech-to-Text Data Collection Pipeline A web application that collects Amharic speech data by displaying random text prompts and storing corresponding audio-text pairs in AWS S3. Designed to streamline dataset collection for NLP model training. --- ## 🚀 Features - 📝 Displays Amharic text for users to read aloud - 🎙️ Users upload recorded audio of the prompt - ☁️ Audio + text stored in an S3 data lake - 📊 Scalable for multilingual voice dataset creation - 🧩 Built with Flask, Boto3, HTML/CSS --- ## 📁 Project Structure speech-to-text-pipeline/ ├── app/ │ ├── app.py │ └── templates/index.html ├── data/sample_texts.txt ├── utils/s3_utils.py ├── uploads/ ├── requirements.txt ├── .gitignore └── README.md --- ## 🛠️ Tech Stack - **Backend**: Python, Flask - **Cloud**: AWS S3 via Boto3 - **Frontend**: HTML (Jinja2 templating) - **Deployment Ready**: Can be hosted on Render or Heroku --- ## 🧪 How to Run Locally ```bash git clone github.com cd speech-to-text-pipeline python3 -m venv venv source venv/bin/activate pip install -r requirements.txt # Add your AWS credentials via `aws configure` python app/app.py Then visit 127.0.0.1 in your browser. 📬 Contact Built by Sayli Nadkar LinkedIn linkedin.com GitHub github.com --- After pasting, **save the file**, then push it: ```bash git add README.md git commit -m "Add project README" git push