video-compressor

Run the compress script with the input path: python3 scripts/compress.py -i <file>. It uses CRF 23 with a 2500k maxrate cap by default, which adapts bitrate to scene complexity and works well for screen recordings and courseware.

Use the trim_silences.py script with the video path. It detects segments that are both audio-silent and visually static, outputs a trimmed MP4, and saves the removed clips plus a JSON timestamp report for review.

Yes. The tool auto-detects Apple Silicon and uses hevc_videotoolbox hardware encoding, achieving roughly 2-5x realtime for 1080p60 content. You can also force encoders with --codec hevc_vt, h264_vt, x264, or x265.

This happens when brew upgrades a dependency like x265 without relinking ffmpeg. The script runs a smoke test that detects this crash and prints the fix: run brew upgrade ffmpeg or brew reinstall ffmpeg.

The tool accepts mp4, mov, avi, mkv, webm, flv, wmv, and ts files. You can pass multiple files, directories, or a mix, and it processes them concurrently with a configurable worker count.

It is not suited for video editing, audio extraction, or format conversion tasks. It always outputs MP4 with AAC audio and never overwrites originals, so use a dedicated editor for cutting creative content.