Abstractive Summarization of YouTube Videos Using LaMini-Flan-T5 LLM

In today's era, YouTube is considered as one of the best places to look for quality videos. Videos about ant topics can be found in YouTube. But the large volume of content in YouTube can be overwhelming for some people. This could be due to two reasons: people find that watching length videos are inefficient and selective the right video to watch. To tackle this problem, we have devised a solution called “YouTube Summarizer” built using artificial intelligence technology. YouTube Summarizer is an AI web application built using large language models and trained on large corpus to imitate human-like language. It works by providing a detailed summary of a YouTube video using LaMini-Flan-T5 model, which is a fine-tuned version of Google's Flan T5 LLM. The aim is to get the transcript of the YouTube video by retrieving the Video's ID by the YouTube API, tokenize the text using tokenization modules, loading a LLM and integrating using Hugging Face pipeline to summarize the text. We then use Text-To-Speech API to convert the generated summarized text into audio. This entire application is wrapped by Streamlit interface.