Deep Generative Multimedia Children's Literature

TL;DR

This paper presents a system that uses pretrained deep neural networks to create multimedia children's literature, demonstrating the potential and challenges of AI-generated creative content across multiple media domains.

Contribution

It introduces a novel approach for generating multimedia children's stories using only publicly available pretrained models, highlighting cross-domain creative AI applications.

Findings

Generated stories showcase diverse multimedia content.

The system demonstrates feasibility of AI in children's literature.

Identifies challenges in AI-driven creative multimedia work.

Abstract

Artistic work leveraging Machine Learning techniques is an increasingly popular endeavour for those with a creative lean. However, most work is done in a single domain: text, images, music, etc. In this work, I design a system for a machine learning created multimedia experience, specifically in the genre of children's literature. We detail the process for exclusively using publicly available pretrained deep neural network based models, I present multiple examples of the work my system creates, and I explore the problems associated in this area of creative work.