YouTube Keywords Mention Tracker with Whisper (deployed on Streamlit)

[dimitryzub]

I've built a fun demo project that uses whisper to transcribe incoming videos from pytube (even if no captions are available) and use transcribe data to find target keywords and return YouTube timestamp URL(s) where the keyword appears.

The main idea is taken from Mention which tracks mentions across different websites (twitter, StackOverflow, etc) but for videos only.

Scenario: user wants to track someone said certain keywords like "Serp Api" in a given search query(ies). Or a query "switching from iphone to..." and target keywords like "poor" (for poor quality), or "No..." (no support for certain feature that android has) and similar keywords that may identify reasons of switching.

streamlit demo | dimitryzub-youtube-mention-tracker repo

This is how transcribed results look like with a target keyword "serp api" to look in videos:

youtube-videos-tracker.mp4

Improvements:

1. speed up transcription. It takes a lot of time.