Розробка інтелектуального веб-додатку мультимодальної адаптації текстового контенту у відеоформат

Abstract

This work presents ReadNet, an intelligent web application for multimodal adaptation of text content into video format. The system transforms books into immersive video experiences using AI-powered content reformatting via Large Language Models, text-to-speech synthesis with word-level timing synchronization, dynamic image display, and background audiovisual content. The platform is built with Next.js (React, TypeScript), PostgreSQL, and AWS S3, integrating LLM APIs through LiteLLM and Microsoft Edge TTS for narration. Personalization is achieved through customizable reading modes and user interest profiles that guide AI content transformation. The system supports PDF book upload, automatic text and image extraction, content caching, and seamless page-to-page transitions with preloading. Video mode synchronizes background video, TTS narration, word-by-word subtitles, and book illustrations into a single cinematic experience