(Reprinted from the SBE’s newsletter, The Signal )
By Vicki Kipp, CPBE, ATSC3, CBNT
Have you ever watched a technical video that is timely, but difficult or impossible to understand? There are plenty of reasons to use closed captioning and not just for the hard-of-hearing community. Perhaps the audio is low or dialog is drowned out; the audio is hissy, overdriven, or cuts in and out; the audio is overpowered by nearby mechanical noises; nearby people or pets – yours or the presenter’s – are being loud; you’re in a quiet space where you can’t play audio aloud; the presenter is clear, but you can’t hear comments from the unmiked audience; you retain written information better than spoken information; you absorb information best when you hear and read information at the same time; you want to cue the video to a specific moment; you only use captions when you’re watching speakers who have an accent or speak a different languages e.g., Game of Thrones; or when you’re learning a new language like IP networking, AoIP, A/300, coding, or FCC-ese.
The reality is that closed captions increase engagement. Adding closed captions to a broadcast engineering video can make the dif- ference between that video getting very few views and that video being appreciated and recommended to others.
In addition, adding captions brings secondary benefits. If a transcript of the closed captions is available, you can skim it for rele- vance before watching and use it in place of taking notes. Search engines index closed captions to use as Search Engine Optimization (SEO) keywords, increasing discoverability.
Social media platforms play videos with the audio turned off by de- fault, making it hard to draw in viewers without captions. If you mon- etize your social media videos or seek views, user-provided closed captions result in more views (reach) and longer views (completion rate). Search engines and social media give a higher rank and more prominent placement to videos with user-generated captions. This ranking does not apply to auto captioned videos.
Closed captioning, which was developed for people who are deaf or hard of hearing, benefits all of us. Captions will enhance your video’s value. So how can you caption your non-broadcast videos?
Caption Styles
You should first know that there are two caption styles: roll-up and pop-on.
Live stenocaptions and Automatic Speech Recognition (ASR)/ Speech-to-Text (STT) AI captions are roll-up style. Roll-up captions have two or three vertical lines that scroll up one line at a time, and the captions lag the live audio. Pop-on captions, meanwhile, are composed after production ends.
Pop-on captions are displayed as a one- to three-line block. A pop- on caption block is switched out for the next caption block, in sync with the audio.
There are options to auto-caption a video. You can allow YouTube, Facebook, Twitter, Instagram, and TikTok to generate free auto captions. They may not be perfect, but something is better than nothing. For verbatim captions created by a human (not autocaptions), Rev is an option for bargain captioning and CaptionMax is a choice for broadcast-quality captioning.
DIY Captions
If you want to do it yourself, you’ll first need a transcript. Manually transcribe your video or use ASR transcription. Different platforms use different speech engines, and their accuracy varies. Otter.ai offers limited free auto transcription. Or you could host a meeting on Zoom, screen share your video and audio, turn on Zoom captioning, and harvest the transcript.
YouTube has the easiest auto caption process. Upload a video to YouTube Studio. Set visibility to Private, Unlisted, or Public and select the language spoken. This works if the audio is clear and can take minutes or hours to generate auto captions. If the audio is noisy or has singing, YouTube may not be able to auto caption it.
Once you have the captions, I suggest proofreading everything. If you Google “auto caption fail” you’ll see why.
Note: The FCC regulates broadcast captions for accuracy, completeness, positioning, and synchronicity. The agency will issue fines. The Described and Captioned Media Program’s (DCMP) Captioning Key is the gold standard for best practices.
Know that ASR struggles with homophones. When proofing, correct “sound-alike” words. Also, captions should be verbatim and not paraphrased. When possible, take the extra step to describe non- speech sounds. Some examples:
[folksy music]
[mooing]
[1 kHz tone]
You can edit the captions online using YouTube‘s Subtitle editor or download a text or captions file to edit them locally. To edit in a word processor, download a transcript without timecode. On YouTube, click on the three dots button under the video to view the transcript. Select “Show transcript.” At the top of the transcript, click on the vertical three dots button to toggle the timecode off.
Select the entire transcript, copy, and paste it into a document. Run spelling and grammar checks. Auto captions struggle to know where sentences begin and end. Revise punctuation and capitalization. Multiple languages, crosstalk, fast talk, or mumbling result in missing words. Auto captions may repeat words or phrases multiple times.
Indicate a speaker change by inserting double chevrons (>>). You can also show a speaker change with a dash and a space (- ).
For more readable pop-on captions, hit carriage return <CR> at the end of a sentence to start a new caption block. Other-wise, YouTube will display your transcript as word-wrapped roll-up captions. Here’s an example from an SBE WEBxtra:
>> Chriss: Good afternoon. <CR>
>> Kirk: Hi, Chriss. <CR>
Our guest is Andrea. <CR>
>> Andrea: Hi. <CR>
When you are finished editing, save the transcript as a plain text UTF-8 file to remove incompatible characters. Upload the text file to YouTube subtitles. YouTube syncs your transcript with the audio, timestamping individual caption blocks. Once complete, tweak YouTube’s timing if needed. Click Publish.
If you download a caption file, you can edit it with free software such as Nikse.dk Subtitle Edit, NCAM CADET, or paid software including Synchrimedia MovieCaptioner, Adobe Premiere Pro, Telestream CaptionMaker, EZ Titles.
Subtitle Edit (SE) imports/exports most caption formats and plain text transcripts. Build a dictionary to replace oft-misspelled words with SE’s Multiple Replace. Some possible additions: six- pack=fixed path; iPod/pause=IPAWS; Nasdaq=DASDEC. Upload the corrected caption file to YouTube to replace auto captions.
Caption File Formats
While broadcast caption files are embedded in the video, Inter- net captions are sidecar files that get uploaded after the video. IP platforms use many caption formats. The hexadecimal SCC Line 21 caption format carries a timed transcript with formatting and positioning. DXFP, SRT, SVB, and VTT captions are human-readable.
Take the time to make your videos accessible, discoverable, and free of awkward auto-caption fails.
Leave a Reply