Live captions in a room or other physical venue do one important job - and how well they do that one job is therefore critical. Your audience, preciously and expensively acquired, can follow what is being said as it's said. Hybrid events have other audiences, though. There is the person in the third row of a breakout room who missed a figure because the speaker turned away from the microphone. The person who couldn't get into the room which was oversubscribed. The person who came in late. There is the colleague in Singapore who joins the town hall 20 minutes late. There is the attendee whose first language is not English and who wants to reread a sentence before the next one arrives. The more people you have, in the more locations, the more chance you will not just have edge cases but your edge cases become your audience engagement profile.
Captions scroll past and disappear - great for those people paying attention - less so for those who aren't (for whatever reason). Timbra, CaptionHub's real-time captioning and translation platform for live streams, in-room events and webinars, gives your audience two ways to catch up. A live transcript for events stays on the page, so people can scroll back, search for a name or a number, and read in their own language, and then a live AI summary that gives anyone who stepped out or arrived late a short recap of the session so far.
Who live captions leave behind
Captions sit in two lines at the bottom of a video player and last a few seconds each. That works for keeping up and works badly for catching up. If you missed a sentence, it's gone.
Plenty of people in any audience need more than that. The World Health Organization estimates that more than 1.5 billion people, about one in five worldwide, live with some degree of hearing loss, and 430 million of them have disabling hearing loss. Global events also bring in attendees who are following in a second language. One caption track in one language serves only part of that room.
Events are also changing in their nature. Sit in a talk at re:Invent and you're just as likely to be having a side conversation in the same room as the speaker.
People also miss different amounts. Someone who lost one sentence wants the exact words. Someone who arrived 20 minutes in wants to know what happened, and scrolling through 20 minutes of transcript on a phone while the speaker carries on is a lot to ask.
The live transcript gives people the detail
A live transcript is the running text of the session, building up on screen as people speak. Because it stays visible, attendees can:
- scroll back to a point they missed without interrupting anyone
- search the page with their browser's find tool for a speaker name, a product or a figure
- switch to another language and read at their own pace
Timbra's Live transcript widget does this. It shows a scrollable transcript of the live stream with a language selector, so each viewer picks the language they want to read. You add it to any web page with a short JavaScript snippet, and the embed code sits ready in the Widgets tab of the live project.
The live AI summary gives them the gist
An AI summary condenses what has been said so far into a short recap. Timbra can generate one while the event is running. Set it to generate automatically and it refreshes every minute, so the recap keeps pace with the session. Or generate it manually at the moments you choose, such as the start of a break.
Attendees find the latest live AI summary under Catch up in the in-room companion, the mobile web page they open by scanning a QR code in the venue or following a link you share. A late arrival can read the recap, then switch to the live text and follow along from there.
Summaries help attendees get more from the whole programme
Each room's Timbra project has its own companion link and QR code, and attendees don't have to be in the room to open it on their own devices. Put the links for every room in your event app or on the agenda page, and the summaries start working across the whole conference.
- Overflow rooms. When a popular session fills up, the people left outside, some of whom have flown in for it, can follow the live summary and transcript from their phone in the corridor or an overflow space instead of missing the session.
- Choosing between sessions. Two talks clash. An attendee can open both summaries, see which discussion is more useful to them, and pick.
- Switching mid-session. Someone in a session that isn't what they hoped can check what is being discussed next door and move, already knowing what they've missed.
Attendees spend more of the day in sessions that matter to them, and get more value from the ticket and the trip.
Where they fit in your event plan
A few places they fit well:
- The event page (or any digital page/site). Put the transcript next to or under the stream so remote attendees can read along, and share the companion link so people who join late can read the recap first. The transcript widget is a simple piece of JavaScript code like this:
<div id="live-transcript-widget-vue"></div>
<script src="https://cdn.captionhub.com/live/transcript-widget.js"></script>
<script>
document.addEventListener('DOMContentLoaded', (_event) => {
window.liveTranscriptWidget({ id: "example-id" });
});
</script>Live transcript widget setup guide
- Breakout rooms without an AV crew. Stream the room audio to Timbra and show the transcript on a screen or a laptop. People who are hard of hearing can follow the session, and anyone who arrives from an overrunning session can scan the QR code and catch up.
- The main stage. Timbra's in-room display puts captions on the venue screens, and the companion lets attendees follow on their own phones in their chosen language, with the summary waiting for anyone who stepped out to take a call.
- Webinars and town halls on Microsoft Teams. Teams sends the event audio to Timbra, and Timbra's webinar widget shows the running transcript, with translations, on the page alongside the stream.
You do not need to choose. Run captions in the player, a transcript for people who want to scroll and search, and a summary for people who need the short version.
Accuracy is key: get the words right before you go live
People only rely on a transcript or a summary if they trust it, and the errors that hurt most are the ones everyone notices: the CEO's name, a product name, an acronym the whole company uses. The summary is built from the transcript, so a name the transcript gets wrong can turn up in the recap as well. Fix those before the event.
- Build a custom dictionary and term base. Add speaker names, product names, internal acronyms and anything with an unusual spelling. Timbra uses them during transcription and translation.
- Set up the AI summary. Switch on your OpenAI or AWS Bedrock connector, choose automatic or manual generation in the AI summary tab, and add an event description and speaker information so the summary has context.
- Brief your speakers. Ask them to stay close to the microphone, avoid talking over each other in panels, and spell out new names once.
- Pick your translation engines. Timbra works with Amazon Translate, DeepL and Google Translate, so you can test which reads best for your language pairs before the day.
- Run a test stream. Check the transcript and the companion on the devices people will actually use, including a phone and the venue screen.
On the day, speaker detection breaks the text up by who is talking, and audio event captions mark applause, laughter and music. Timbra also supports primary and secondary transcription, so if one fails the other carries on and viewers do not have to rejoin.
After the event
Once the stream ends, you can download VTT files of the captions in every language from the project's History tab, so you have a record of what was said for QA and auditing. Analytics show how many viewers turned captions on and which languages they picked. That language data is useful when you plan the next event, because it shows which languages your audience actually used.
Plan it into your next event
Captions help people keep up with a live stream. A live transcript and an AI summary help the people captions miss: attendees in the room, people who join late or end up outside a full session, and anyone reading in a second language. Timbra runs captions, the transcript and the summary from the same live project, with no on-site captioners, no hardware being shipped, and no extra tools to source. If you are planning a hybrid conference, town hall or series of breakout sessions, add them to the run sheet at the same point as the stream.
Already using Timbra? The Live transcript widget guide and AI summary guide show you how to add both to your next live project. New to Timbra? Book a Timbra demo and we'll show you the Live transcript widget and a live AI summary running on a stream.
See CaptionHub in action
Get a personalised walkthrough of captioning, translation and voiceover at enterprise scale.
