Shorts Generator

This tool will automatically generate short form content in the form of text-based stories spoken over stimulating background videos.

Example Video

video.mp4

Installation Instructions

git clone https://github.com/aj-arts/shortsGenerator.git

cd shortsGenerator

Create a virtual environment

python3 -m venv venv

Activate venv and install dependencies

source venv/bin/activate

pip install -r requirements.txt

IMPORTANT

To use captions on videos with MoviePy, you MUST download a program called ImageMagick

You can find the download links for Linux, Mac, and Windows here

In order for the generator to scrape Reddit stories or use ChatGPT you MUST create a .env file with

OPENAI_API_KEY = "[your openAI api key]"
REDDIT_CLIENT_ID = "[your reddit client_ID]"
REDDIT_CLIENT_SECRET = "[your reddit client_secret]"
REDDIT_USER_AGENT = "[your reddit user_age]"

Without this you cannot access the Reddit or OpenAI API!

Usage

To use the tool run the command

python generator.py

BE SPECIFIC WHEN PROMPTED FOR VIDEO BACKGROUND

For example:

subway surfer background gameplay for tiktok

short nature background video relaxing

WHEN USING CHATGPT

With your prompt to chatGPT make sure to include a word limit, it may also help to be specific. We recommend not to go over 750 words. Using over 500 will most likely require you to increase the number of tokens in the generate_req() function on line 28 in generator.py.

See the README in content for more information.

Examples

generate a random AITA reddit story with a word limit of 150

generate an AskReddit thread with responses no longer than 200 words

generate a cute story about someone getting into a car crash and falling in love that is under 300 words

Additional Notes

You may come across an error such as

Unexpected renderer encountered.
Renderer name: dict_keys(['reelShelfRenderer'])
Search term: tetris
Please open an issue at https://github.com/pytube/pytube/issues and provide this log output.

This is an issue with pytube and hopefully will be resolved soon. It does not impact the performance or outcome of the video, but if you wish to mute the issue temporarily navigate to

venv/lib/python3.X/site-packages/pytube/contrib/search.py

and after line 150 add

if 'reelShelfRenderer' in video_details:
    continue
            
if 'showingResultsForRenderer' in video_details:
    continue

We use a graveyard in order to prevent reuse of Reddit posts and YouTube videos. See the content README and video README for more information.

Future improvements

Work towards integration with uploading to multiple media platforms - Youtube, Tiktok, Instagram
Provide more customizability for video options
Improve TTS engine to be more realistic

Written by:

Sarvesh Thiruppathi Ahila - text scraping

Ajinkya Gokule - video processing and captioning

David Gesl - text to speech

Colin Pannikkat - video scraping, documentation

rankuptuts/shortsGenerator