I’m thinking of an idea: to develop a software program that takes a podcast that is 1 hour long and trims it automatically. Then, creates Youtube Shorts, i.e., small duration clips from the long video through machine learning and AI I am guessing; I’m not so good at programming.
Would love to have a meaningful discussion on this idea and possible steps to come up with a possible software…
1 Like
I cannot help how to do this project, but I know of one multi-media library application where you can upload video clips or long videos, annotate them, tag them, and then perform a search on the videos. The result of the search is a new video that will collect the clips from multiple videos, long and short. Though this process is not automated, it helps you to figure out the algorithm and the techniques as well as the Python libraries used to build this feature.
The application is called pandora, written in Python.
An implementation of this software can be seen in two massive video libraries:
and
1 Like
Not sure what is the goal, but ffmpeg allows one to clip any video, separate a/v, change the codec etc.
ffmpeg -ss -t -i -acodec <audio codec to use, "copy " keyword will use same codec as source> -vcodec <video codec to use, "copy " keyword will use same codec as source>
Here is a typical command I use very regularly
ffmpeg -ss 452 ffmpeg -ss 452 -t 146 -i some-youtube-video.mp4 -acodec copy -vcodec copy /tmp/NoJunkYoutubeVideo.mp4
vlc too has innumerable cli parameters to do the above, as well as ability to extract a clip using the VLC GUI .
ffmpeg cheat sheet.