Video editor

The video editor takes the raw recorded files from the broadcaster, processes them, and uploads them to YouTube (or whatever).

Overall priorities

  1. No learner (or anyone not staff) video, audio, names, etc. are present in the recordings.

  2. Good descriptions.

  3. Removing breaks and other dead time.

  4. Splitting videos into useful chunks (e.g. per-episode), perhaps equal with the next one:

  5. Good Table of Contents information so learners can jump to the right spots (this also helps with “good description”.)

Modern: livestream method

See also

The train-the-trainer course has a page giving a tutorial on video processing.

Modern livestream courses produce videos without any learners in them. In this case, using https://github.com/coderefinery/ffmpeg-editlist is sufficient. Look at that repo for instructions. As an example, check out https://github.com/AaltoSciComp/video-editlists-asc for some past workshops. For example, kickstart-2022-winter.yaml is a reasonable starting point to copy.

It’s our standard to have these videos on YouTube by the same evening the course is held. It may be hard, but it’s better to reduce the quality to make it happen quickly than wait a while to get it perfect (otherwise it might not happen at all).

The steps you should take:

  1. Decide the total amount of time you will spend. Probably no more than 2hr per half-day of teaching.

  2. Open the .srt file. (it’s a text file, but using a subtitle editor is probably worth it)

    • It is usually in the video-processing repository.

    • You don’t need to make it look perfect. It should look useful. Grammar/spelling doesn’t matter that much. The times when Whisper stops including punctuation don’t matter so much.

    • I usually correct unix commands (search dash, hyphen, minus, dollar, space) for example. A learner is likely to try to read or copy these from the subtitles, even if they are in the video itself.

    • All names should be replaced with [name] (including instructor names). This is probably done before you got the .srt file, but you can check.

    • Push it to git.

  3. Create the editlist in the .yaml file.

    • The .yaml file should be pre-created and is usually in the course website repository or video-processing repository.

    • Make sure the session is there and has proper descriptions

    • Set the overall cut points for each session, break, and exercise (these are relatively easy to find)

    • Set timestamps for each new episode/section

    • Set timestamps for other important things learners may want to quickly find (important configuration, introducing exercises, etc.)

  4. If you want, you can try generating the final videos yourself. But the video uploader is likely to do this themselves anyway. Some commands below:

    First, arrange files: compared to $PWD, raw files are in raw/*{.mkv,.srt}, and output gets written to out/. The .yaml editlist file can be in $PWD or somewhere else.

    First, run in check mode (-c) and limit to outputs matching day1 (-l). This will report any major errors. If this passes, it’s good enough For me to take over. Look at .info.txt files to see if the description looks OK.

    $ ffmpeg-editlist ~/git/2025-03-25-workshop/video-editlist.yaml $PWD -c -l day1 --srt

    You can run and generate .mkv files, but don’t re-encode (first few seconds are messed up). You can see if .mkv looks roughly correct:

    $ ffmpeg-editlist ~/git/2025-03-25-workshop/video-editlist.yaml $PWD -l day1 --srt

    When it’s all done, run with --reencode to create final files (you don’t need to upload, video producer will regen):

    $ ffmpeg-editlist ~/git/2025-03-25-workshop/video-editlist.yaml $PWD -l day5 --srt --reencode

  5. Push the .yaml to git and push the new .srt file via git.

    • Tell the main video manager it’s done and they will check, encode, and upload to the respective places.

If the learner Zoom is recorded

If learners may be in the recordings, they need detailed checking before they can be posted. See Video checking OLD for the preparation work and Video editing OLD for the processing work.

In practice, if things are recorded this way, they are almost never released because it is too much work and it never gets done.

Editing hints

  • https://mpv.io is a command-line video player. - Hotkey O (capital letter O) to turn on time display

    permanently

    • Seek with left/right (~10 sec) or up/down (~ 1 minute)

      • Run with --hr-seek to seek exactly, not to keyframes

    • https://github.com/Kr4is/mpv-copy-time - mpv script that lets you copy video times with a hotkey. You might want to edit the hotkey by editing the script.

  • How to run ffmpeg-editlist: - first with –check - then with no other - then –reencode