Audio and video

Teaches agents to work with audio and video files using ffmpeg (which must be installed on this computer) and gives them yt-dlp in an environment of its own, for downloading media you have the right to download.

The skill media-tools covers converting, cutting, joining, compressing, choosing tracks, extracting audio, GIFs, thumbnails and contact sheets, subtitles, vertical versions for social media, and checking every result. For longer edits it renders a cut list (the parts to keep, or to take out) in one go, also from a transcript. For podcasts and voice recordings it measures loudness, peaks, clipping, background noise and pauses, levels the sound to a target such as -16 LUFS, and shortens long pauses. Two small scripts do the error-prone parts; they need nothing beyond ffmpeg and Python.

Other plugins build on this one (Speech to text).