How to help with transcribing

[nedtwigg]

This video shows how to do it. It's 14 minutes long, but the first 4 minutes just pitch MyTake.org in general - you can skip to 4:00 and just watch the last 10 minutes if all you care about is the technical how-to.

If you want to do this work, you'll pick a video, crunch on it for 1-4 hours, and then you'll be done! You'll learn a lot about what the hot topics were at that point in time, and you'll learn even more about the ambiguities in human language - how we naturally hear pronouns and filler words even if they aren't actually enunciated, and lots of other interesting quirks. It's fun!

Here's how to do it:

• open a terminal (cmd on windows, terminal on mac/linux)
• run git clone https://github.com/mytakedotorg/mytakedotorg
  • if it fails, you need to install git: https://git-scm.com/downloads
• that will create a new mytakedotorg directory, which you can cd into
• then run ./gradlew transcriptGui (on windows, leave off the ./)
  • if it fails, you probably need to install JDK 8

If you get that far, then you will be able to see which videos are ready to be transcribed. If you want to know how long it will take, it's pretty conservative to just double the running time - e.g. a 1 hr debate will take ~2hrs to fixup the transcript. Call Ned before you get too invested, and he will make sure you're not working on the same one as someone else (by updating the "Dibs" column in this spreadsheet).

The GUI is saving as you go, so you don't have to worry much about losing your work. If you already know git, feel free to make commits as you go and push up a PR. If you don't know git, you don't have to learn it, just do this:

• if you run git status, it will show you the file you have been changing
• just email that file to Ned, and he can put it into the website for you

And that's all there is to it! Three major caveats:

• if you ever get stuck, don't spend time trying to fix it, just call Ned. If things are working right it's really simple, just like the video above. If anything is weird or confusing, it's probably Ned's fault, so you should just call him.
• if the video ever skips, loses audio, anything like that, stop and call Ned.
  • it probably means we have to find a new source, and we'll probably have to start over for that video
  • very sorry, if it's any consolation Ned and Todd have had to start over a few times
  • and from those wounds, we've finally procured a very reliable source of raw video, and we're pretty sure it won't happen again for any of the remaining videos, but be warned! It doesn't do any good to steamroll ahead
  • definitely call Ned and tell him, maybe he can finagle something to salvage it, and he'll definitely try, but probably it will be quicker to start over
• if you get confused about a number (e.g. $1,000 vs one-thousand dollars vs 1000 dollars), here are the guidelines:
  • the guiding principle is that it's a transcript, so the goal is to match the sounds
  • 50 percent, not 50%
  • 950 dollars not $950
  • one trillion not 1,000,000,000,000
  • we can (and probably will) change the exact details of how we represent numbers at a later date, so don't worry too much about it
  • if you're really curious, this is a deep-dive

[nedtwigg]

As of Mar 31 2020, this is broken on Windows (Mac & Linux only) - will have a fix soon!

[nedtwigg]

Fixed for windows users as of Apr 2 2020. If you were having problems, run git pull and that will install the fix.

[nedtwigg]

@ftwigg is our first volunteer, and he raised a great point which is worth documenting here.

People naturally pause and um, ah, er, etc. The YouTube transcript usually does not have these ums, and the Newspaper often does, but not always. Our goal is for the transcript to match what was spoken as exactly as possible. That means that we want to keep the ums.

There is also the matter of word- word and word - word. The hyphen-pause should have a space on each side word - word, not word- word. It is also okay to use commas. I need, uh, a small pause. But this idea - a big idea - it needs a big pause. It's your call!

Minor caveat: You may find that you hear an "um" which is in neither the Newspaper nor the YouTube. Most such "ums" will probably not be found - the tooling doesn't help at all, so it's just luck if you find one. Ideally we would get them all, but for now it's okay if we miss them, so don't worry about it.