An AI portal that turns books and podcasts about Troy into short narrated videos, letting students watch the city’s history unfold—on a platform that can later scale to any city.

https://huggingface.co/spaces/ankitkupadhyay/RetroVision

We extract data from 3 different sources:

1. A carefully compiled book about the history of Troy(https://archive.org/details/cu31924028826472/page/56/mode/2up)
2. Online text sources
3. Podcasts

To run data extraction, you first need to download the book from https://archive.org/details/cu31924028826472/page/56/mode/2up into the data_extraction directory. This project requires Python version 3.13.

After this, run

cd data_extraction
pip install -r requirements.txt
export GEMINI_API_KEY=<insert your API key here>
python extract_events_by_decade_book.py
python extract_events_by_decade_podcast.py
python extract_events_by_decade_web.py

This will generate csv files with decade based historical information, which is then ingested by the video generation notebook.

All code related to video generation is in the video generation notebook.