Captions are a tool that has been underused for the creation of memes.

What it does

Takes as input a string and using the captions of a bunch of YouTube videos it has downloaded it searches for clips in those videos (or synthesizes a clip of the audio if it's not found) and mashes them together to form a clip with that text.

How I built it

I built using lots of Python scripts and Google Producs (T2S, S2T [eventually], YouTube).

Challenges I ran into

I made the mistake of downloading the video in mp4 format, this made the job of splitting videos much more computationally expensive than I thought. I thought I would be able to pre-split all the videos but I needed to do it on-demand instead to allow for how slow it is (even multi-threaded). If I had picked another format I could do this faster.

Accomplishments that I'm proud of

It kinda works!

What I learned

Video codecs are difficult. I can think of plenty of clever ways to optimize what I've done here.

What's next for Video Masher

Probably not going to work on it further but some ideas:

  • Web frontend
  • S2T to split captions on individual words
Share this project: