/Sharetape-Open-Source

Script that takes any long form video or podcast and outputs clips for social media

Primary LanguagePythonMIT LicenseMIT

Sharetape-Open-Source

This script uses NLP to determine the 10 best topics from a podcast or long form video. Then based on these topics it cuts 30-60 second clips for TikTok, Instagram, or YouTube Shorts. You can even crop the video to 9:16 format and add captions if needed.

Demo

Sharetape Open Source Demo

Install homebrew dependencies

$ brew install ffmpeg
$ brew install imagemagick

Install requirements

$ python3 -m venv venv
$ source venv/bin/activate
$ pip install -r requirements.txt

Download Vosk Library

This is the language library this speech to text uses. Download this Here

Once downloaded unzip in your project directory.

Execution

There are 4 flags

  • -v or -video is the video path. Only .mov or .mp4
  • -c or -crop crops the video to 9:16 format, default is False
  • -ca or -captions adds captions, default is False
  • -cl or -clipLength is length of clip, default is 30 seconds

Clips will be output in a directory with a unique ID i.e. c088af43-362b-4837-bf29-d9122008f457/clips

Example:

$ python main.py -v 1086final.mov -c True -ca True -cl 60