Summarize-Pro: A Transformer-Based Framework for Multimodal and Multilingual Content Summarization

The rapid expansion of digital content has created a strong need for efficient summarization systems. Summarize-Pro is a multimodal and multilingual platform that supports multiple input formats, including text, documents (PDF, DOCX, TXT), web URLs, and YouTube videos—transcribed using OpenAI Whisper. The framework leverages transformer-based models optimized for different input types: BART-Large-CNN for documents, Pegasus-XSUM for web content, Long-T5-TGlobal for long transcripts, and mBART-Large-50 for multilingual summarization. It also integrates auxiliary modules such as KeyBERT for keyword extraction, BERTopic for topic modeling, RoBERTa-SQuAD2 for interactive question answering, and gTTS for generating audio summaries. Users can further customize summary style and prompts. Natural Language Processing (NLP) techniques are combined with multimodal and multilingual capabilities, Summarize-Pro produces concise, contextually rich, and accessible summaries that enhance comprehension and user engagement across diverse digital content sources. Evaluations are done using ROUGE metrics which demonstrate high coherence making Summarize-Pro a scalable, user-centric tool for global content understanding.