Webinar captions and subtitles that people can read

Some of your attendees will watch with the sound off. They’re on a train, in an open-plan office or next to a sleeping baby. Some are deaf or hard of hearing, and others follow written English more easily than spoken English. Webinar captions bring all of these people with you. This guide covers the different kinds of captions, the usual file formats, how to check automatically generated text and the rules of thumb that keep captions readable.

Who captions help

Deaf and hard-of-hearing people are the obvious group, and for them captions decide whether they can follow the webinar at all. Then there are the many attendees who simply can’t use sound at that moment: commuting, at their desk, waiting for an appointment.

A third group tends to be overlooked. People who speak English as a second language often read technical terms, numbers and names more easily than they catch them by ear. Native speakers benefit too whenever a price, a web address or a product name comes up. Once it’s on screen as text, nobody has to rewind or ask in the chat.

A note on the terms

The terms get used loosely, so a quick note. In the US, “captions” usually means same-language text that also describes relevant sounds, while “subtitles” means translated dialogue. In the UK, “subtitles” covers both, and the fuller version is often called subtitles for the deaf and hard of hearing. This guide uses “captions” for same-language text and “subtitles” for translations.

Open and closed captions

There are two basic forms, and they behave very differently in practice.

For webinars, closed captions are almost always the better choice. You can fix mistakes in minutes, attendees switch them on when they need them, and a new language is just another file. Open captions make sense for short clips on social media, where videos often autoplay silently and the player won’t show a caption file.

File formats such as WebVTT and SRT

Closed captions live in a plain text file alongside the video or audio. Two formats come up most often.

WebVTT (.vtt) is the web standard. Browsers display WebVTT files directly alongside video, and many webinar players read the format too. A file looks like this:

WEBVTT

00:00:01.000 --> 00:00:04.500
Welcome to the webinar.
Thanks for joining.

00:00:05.000 --> 00:00:08.200
Over the next 30 minutes
we'll answer three questions.

SRT (.srt) is older and very widespread, especially in editing software and on video platforms. Each block is numbered, and milliseconds are separated with a comma rather than a full stop. Most tools convert between SRT and WebVTT without any manual work.

Save files as UTF-8. Otherwise accented names and special characters turn into gibberish.

Three ways to produce captions

From the finished script

An automated webinar often has a big advantage here: the spoken text already exists. If a synthetic voice reads your script, as described in the guide to AI voices for webinars, every word is known in advance. Alignment tools then only have to match that text to the right moments in the audio. That’s more accurate than any speech recognition, because nothing has to be guessed.

There’s still some work. You need to split the text into sensible chunks and check the timings against the audio. But because an automated webinar plays the same way at every session, you only do it once.

With automatic speech recognition

For unscripted webinars or recordings without a script, speech recognition produces the text. It’s quick, and with clear speech the result is often usable. It isn’t finished, though.

By hand or through a captioning service

For important webinars, poor audio or lots of jargon, manual work can be worth paying for. Professional captioners know the conventions for line breaks and reading speed and deliver files in all the common formats.

Checking automatic captions

Speech recognition makes predictable mistakes. Knowing where to look makes them quick to find.

• Names and product names. An unusual name often turns into a similar-sounding everyday word.
• Numbers. “Fifteen” and “fifty” sound close. Check every price, date and percentage.
• Jargon. Industry terms and abbreviations are frequently misspelt or split into separate words.
• Homophones. “Their”, “there” and “they’re” are classic traps, as are “affect” and “effect”.
• Punctuation. Missing commas and full stops make long captions hard to read.
• Text out of nowhere. During music or silence, some systems produce sentences nobody said. It’s similar to hallucinations in language models, and you’ll only spot it if you check those exact passages.

The most reliable check is to watch the whole webinar with captions on while listening. Note the timestamps of any errors and fix them in the file in one go.

Rules of thumb for readable captions

Captions should be readable at a glance, so attendees don’t lose track of the slides. These values work well as rules of thumb. They’re guidelines, not fixed standards.

• Two lines at most per caption.
• Around 40 characters per line, preferably fewer.
• A reading speed of roughly 15 characters per second. A two-line caption with 35 characters per line should then stay up for a little over four and a half seconds.
• Display time of at least about one second and no more than about six or seven.
• Break lines at natural phrases. Don’t separate an article from its noun or a number from its unit. “Over the next 30 minutes / we’ll answer three questions” reads better than “Over the next 30 / minutes we’ll answer three questions”.

If the voice talks faster than the reading speed allows, you can drop filler words or trim sentences slightly. Never shorten numbers, prices or names.

For display, light sans-serif text on a semi-transparent dark background stays legible on any slide. Keep the bottom of your slides free of important content so captions don’t cover anything. A player that lets attendees change the text size is a bonus.

Captions for deaf and hard-of-hearing attendees

Dialogue-only text reproduces the spoken words. Fuller captions go further. They also describe sounds and music that matter for understanding, such as “[applause]” or “[calm music]”, and mark changes of speaker when more than one person talks.

When a WebVTT file is added to a web page, HTML distinguishes between the kind “subtitles” for spoken text, typically a translation, and “captions” for the fuller form with sounds. In a webinar with a single voice and little music the difference is small. If the presenter announces something audible, like a chime at the start, that sound belongs in the captions.

Captions alone don’t make a webinar fully accessible. Anything that only appears on a slide should also be said, so blind attendees don’t miss it. That’s a job for the script rather than the caption file.

Subtitles in other languages

Subtitles in a second language are the cheapest way to open a webinar to an international audience. Slides and audio stay the same and only a text file is added. For attendees who partly understand the webinar’s language, that’s often enough.

Translated subtitles need the same review as a translated script. A sentence can grow considerably in the target language and then breaks the line length. The guide to translating a webinar with AI explains how to transfer and check content, and the piece on multilingual webinars covers when a full second-language version pays off.

Checklist before your first session

• Captions in a separate file (WebVTT or SRT), saved as UTF-8.
• Every name, number, price and date checked against the script.
• Music and silent passages checked for invented text.
• No more than two lines per caption, broken at natural phrases.
• Captions don’t cover important slide content.
• Whole webinar watched once with captions and sound, and once on a phone.
• The registration page mentions that captions are available.

Frequently asked questions

Are captions a legal requirement for webinars?

There’s no blanket rule for every webinar. Captions are, however, a core part of making content accessible, and accessibility requirements do apply to certain businesses and services. Either way, captions simply let you reach more people.

Which format should I use?

For webinars that play in a browser, WebVTT is the natural choice because browsers support it directly. SRT is useful if you also upload the file to video platforms or use it in editing software. Converting between the two is easy.

How accurate is automatic speech recognition?

With clear speech and good audio it’s often surprisingly accurate, and noticeably weaker with names, numbers and jargon. Always plan a round of corrections. If you have a script, aligning that text is the better route.

Should the replay have captions too?

Yes. Someone watching a replay or recording has the same needs as at the first session. With closed captions you simply reuse the same file.

Final thought

Captions take some effort once and then help at every single session. For an automated webinar with a finished script the effort is especially small, because the text already exists. Check the numbers and names, keep the reading pace calm, and your webinar works for people who can’t hear it right now.