An automated video clipping pipeline that processes long video files to extract engaging, standalone segments. The tool transcribes the media, uses an AI model to detect key moments and timestamps, crops the footage to a vertical aspect ratio, and burns synced subtitles onto the final video using FFmpeg.
Wiring up the clipping and subtitle engine
Spent most of this block stitching together the video pipeline. The main goal was getting long videos automatically chopped down into clean vertical shorts without having to do manual cuts.Got audio transcription hooked up so it can grab timestamps for interesting parts of the speech, then passed those segments to ffmpeg to crop everything into 9:16. Most of the headaches were around subtitle timing and making sure the text renders clean without drifting out of sync. Tested it on a couple longer test clips and it is finally cutting and burning subtitles properly end to end.
