This app takes in a video link and summarizes its transcript.
YouTube Transcript API, Streamlit (for UI), and the Gemini language model were used in creating this program. Additionally, the app consists of two methods: one is for extracting the transcript using the YouTube API, and the other one for summarizing the transcript (design diagram below).
One valuable improvement could be incorporating a translation capability. Majority of YouTube videos are produced in English. To enable global audiences with access to a variety of video materials outside their regional reach, a translation capability might enhance the value of the app.
My key takeaway from this project was how easy it was to use the Youtube Script API and Gemini Pro model. With the help of just a couple methods, I was able to generate well-structured and insightful script summaries. What was interesting is the variance in the model's output for the same input. As an example, below is the output for this linked video. Not only did the word count vary, but there was a significant difference in the style and layout of the output.