We were greatly fascinated by the rise of AI-generated images and wanted to use this technology to make the tedious and boring task of reading books a little more interesting. Readify redefines reading!
Readify takes a book, a part of a book, or any text, visualizes the content in pictures, and creates beautiful animations from it that you can sit back and enjoy. If you think audiobooks are fun, you will love us.
We built Readify by using basic NLP to recognize sentences, pre-trained models for the generation of images, and a differential morphing technique to create good-looking animations.
We faced several challenges during development. It would be a lot better if we could run it on our own models, but that would require a lot more time and data. Python does not have a good sounding text-to-speech library, and that was a huge struggle to deal with.
Despite the challenges, we are proud to have built the project, making it work as we had planned in less than 24 hours. As it's our first time participating in anything of this sort, it's a great achievement for us.
Through this project, we learnt how Generative Adversarial Networks work to produce images that have never existed before and also learnt a lot about time as well as effort management at a hackathon.
Next for Readify, we aim to add a GUI to make our program exponentially more fun and user-friendly.
Built With: moviepy, nltk, python, tensorflow