StoryTime Magic

StoryTime Magic is a Streamlit storytelling application for children aged 3-10. A user uploads a picture, the app generates an image caption with a Hugging Face BLIP model, expands the caption into a short children's story, and converts the story into playable audio.

Features

• Upload JPG, JPEG, or PNG images.
• Generate an image caption using Salesforce/blip-image-captioning-base.
• Generate a 50-100 word story using distilgpt2.
• Convert the story to speech using gTTS.
• Play and download the generated audio in the Streamlit app.

Models and Libraries

• Streamlit for the web interface.
• Hugging Face Transformers for image captioning and text generation.
• BLIP (Salesforce/blip-image-captioning-base) for image understanding.
• DistilGPT-2 (distilgpt2) for story generation.
• gTTS for text-to-speech audio.

Local Setup

Install dependencies:

pip install -r requirements.txt

Run the app:

streamlit run app.py

Then open the local URL shown in the terminal.

Deployment

The app is designed to be deployed on Streamlit Cloud. The repository should include:

• app.py
• requirements.txt
• README.md

After deployment, submit the Streamlit Cloud app URL with the assignment.

Testing Checklist

• Image upload accepts JPG, JPEG, and PNG files.
• The BLIP model generates a caption for the uploaded image.
• The text-generation model creates a story within the required 50-100 word range.
• The generated story is displayed in the app.
• gTTS converts the story to audio.
• The audio player and download button work in Streamlit.