Member-only story
Summarize Any YouTube Video in Seconds with Python!
Intro
I was thrilled when OpenAI launched its latest tool in December 2022. It was ChatGPT version 3.5, and it immediately crossed my mind why I had not dealt with large language models until now. I was amazed by its capabilities — how fast it is and the wealth of information that it can provide for us. Like we having a friend with an IQ of ~150 in every area of life. I was not fearful of its potential harm to our world; Instead, I focused only on the positive aspects. I felt it was a huge step forward for humanity and it immediately opened many doors for me and others around the world!
A bunch of application ideas popped into my mind instantly. One of them seemed like a life changer for my daily routine. I always listen to podcasts, tech news, and learning materials on YouTube, but what I lack is time itself. Therefore, I constantly seek ways to improve my daily habits and create faster, healthier routines. Despite learning a lot from these videos, they still consume much of my time. Moreover, several videos have clickbait titles, then lack the real value they promise. Finding relevant, up-to-date information is really time consuming. Until now!
“What if… So… Uhm… We have AI generated YouTube transcripts…
And we have a large language model that can summarise texts…
Therefore… Woo!! We can have a YouTube SUMMARISER!”
The concept is simple: obtain summaries of YouTube videos without needing to watch them. Video summaries will save time and answer the question of whether it is worth watching the video or moving on to others. Let’s delve into the details in the following sections.
The YouTube Aggregator Pipeline
The goal is relatively straightforward. We need to get the AI-generated transcripts from YouTube and feed them to the ChatGPT API for summarisation. Let’s elaborate these two steps:
