Thank you for volunteering to help with transcript correction. We're consolidating style guidelines as things come up in one place. Please don't worry about already completed transcripts (checking them twice would be a later project), and use these guidelines going forward from here. We are very grateful for assistance with transcribing content.
Tips
- You can skip the intro & outro clips for the podcast episodes, I already have those transcribed.
- If you wish to, and can process the audio faster than we are speaking, there's the ability to speed up the playback.
- There's keyboard controls for Otter in the help menu, that may make it easier than using the mouse to stop/start playback. It automatically stops when you go to correct a word, and starts from the last correction when you re-start it. It's fairly smart about punctuation and fairly accurate.
- Whatever software you're using for playback likely has keyboard shortcuts for things like play/pause and other features. You also likely can go into settings (YouTube gear menu) or hold-press on the play button and see about changing the playback speed. If someone talks really fast, you can slow them down — if someone's pausing a lot, or speaks slower, you can speed them up.
Correction Style/Choices (consistency between transcripts)
https://dcmp.org/learn/captioningkey has some relevant guidance and below are decisions &/or exceptions to those guidelines as needed.
If something presents a burden for the volunteers, please let Crisses know and they'll help work out a compromise.
The most important principle is to promote understanding for the listener (viewer, reader), without adding biases of the transcriber. With recorded content the listener does have the opportunity to replay, to change the speed of the track, reread part of a transcript, etc. Keeping that in mind, we do not need to be overly concerned with the speed of speech vs text (like counting words per second, etc.).
- Fillers, etc. For reading flow and to make things fair for transcribers, there's some determinations we have to make to not bog people down on either side.
- Leave out filler sounds unless there's a long pause that needs to be filled. [uhm] or [hums] should be sufficient — use a descriptive term ("hums") or an audio gloss ("uhm" "um" etc.). If there is filler sound on the audio track, a deaf listener may be concerned that there is a gap in captioning otherwise.
- Word fillers like "so" "like" "Right?" leave in, unless they're doubled "like…like…" then you can reduce it to one instance for legibility (with the caveats below).
- If we're talking really fast, you can remove (or leave out if the machine transcript omits them) some filler words to increase comprehension & reading speed, but make sure you are not altering the meaning of the sentence.
- Partial words, or false starts, can be separated from the text with an em-dash — (transcripts) or double hyphen -- (closed captions) to show interrupted speech. "When pro-- You know, it's like the world went sideways." We're over 50. Sometimes the nouns just don't come to us. lol
- Also don’t overthink it; strive for accurate transcribing, not editing. People can pause or open the sidebar transcript window on YouTube. If you find yourself editing out a lot of filler words, then you're probably editing out too many filler words. They could have edited those out before publishing — so it's on Crisses.
- Background Noises - only if they're relevant to someone who can hear, or have impact on the goings-on in the recording. So intro [music] yes, a fan in the background no.
- If something like Crisses' phone rings and they comment on it then [spooky ring tone] may be relevant. If they get a notification beep but it has no impact on the content at all (they don't comment about it, and everything goes on uninterrupted), then ignore it.
- As a dog sitter, Crisses may comment on a dog or cat interrupting, or they may pause or restart a sentence because of it. So [dog barking], [dog whining] or [dog tags clinking] etc. just enough to indicate what a hearing listener is aware of relevant to the comments made or flow of speech.
- At least one early episode of the podcast was recorded on the streets of NYC, so the background noises may be relevant for context: [city street sounds] for example. Some episodes were recorded while driving [car driving sounds].
Audio Description Notes