Skip to content
Blink Brief
Language:

Guide · Events and stages

Live captioning for in-person events: CART, AI captions on the screen, and captions on attendees' phones

By Blink Brief · Updated  · How pages are made

If someone who is deaf or hard of hearing asks for captions at your event, book a CART captioner well ahead: a person who types every word as it's said, live, onto a screen. For sessions where nobody has asked for CART, free speech-recognition captions from a laptop can go on the main screen if the microphone feed is clean, and attendees can read the same captions on their own phones from a QR code.

Short answer

  • Someone requested CART: book a human captioner; AI captions are not a substitute.
  • No request, a talk in one language: AI captions on the screen from a laptop, free.
  • A big room or a small screen: the same captions on each attendee's phone, free during the beta.

CART, AI captions, or both

ADA.gov lists "open captioning, closed captioning, real-time captioning" among the aids for effective communication, and defines real-time captioning, or CART, as "a service similar to court reporting in which a transcriber types what is being said at a meeting or event into a computer that projects the words onto a screen." AI captions do the same job by speech recognition, with no person checking the words.

Option Who writes the words Price Use it when
CART, on site or remoteA trained captioner, liveBy the hour; $79–$150 under Washington's college contractsSomeone requested captions, or names, numbers and terms must be right
An AI captioning serviceSpeech recognition, run by the vendorSyncWords $30 for one hour; Wordly from about $1,500 for 10 hoursYou want support, a hybrid stream or translated captions
Free AI captions from a laptopThe browser's or the computer's speech recognitionFreeNo request for CART, one speaker at a time, a clean microphone feed

US dollars, read on 7 November 2026 on Washington SBCTC's vendor page and each vendor's own pricing page.

Who chooses matters more than the price. Churches and other religious entities are exempt from title III of the ADA (III-1.5000 of the Justice Department's technical assistance manual); for everyone else, ADA.gov says title II entities, such as a city or a public school, must give "primary consideration" to the aid "requested by the person who has a communication disability", and title III entities, such as a business or nonprofit hosting a public event, are "encouraged to consult with the person". New York State's office for deaf and hard of hearing people is blunt: "When CART services are requested, auto captions or artificial intelligence-generated captions should never be used."

Booking a CART captioner: what to send them

New York State's office advises booking on-site captioning "well in advance of the event (ideally several weeks or months ahead)". Then send, from the two state guides read for this page:

  • Who speaks: "speakers' names, titles, the agenda for the meeting, any specific acronyms", and the presentations (New York).
  • Everything in writing: "schedule, agenda, PowerPoints, handouts, outlines, names, technical vocabulary, websites" (Wisconsin). If you can't send them ahead, bring them on the day.
  • Sound: "it is best to provide a direct feed through the sound system" (New York).
  • A place to work: a table and chair near the presenter with power, where the seating and cords don't block "the screen, presenters, doors, aisles, or frequently used pathways" (New York). Ask whether you provide "a projector, screen, and/or extension cords" (Wisconsin).
  • For remote CART: the meeting link, sent ahead (New York), and a way for the captioner to hear the room.

AI captions on the screen from a laptop

For a talk where nobody has asked for CART, the free Live Captions tool puts very large captions on the projector from a browser, with no sign-up. Connect the laptop to the projector, pick the language being spoken, press Start captions and allow the microphone, then press Full screen (or F). The text size goes from 24 to 160 px, with a dark or light background and Auto-scroll. It works in Chrome and Edge on a computer, or Chrome on Android; Safari works on some Macs and iPhones, and Firefox doesn't.

On the screen, the captions are in the language being spoken; they aren't translated there. Blink Brief never receives the audio, but the browser may send it to its maker's speech service (Google for Chrome, Microsoft for Edge, Apple for Safari) to turn it into text. The Mid-Atlantic ADA Center's events guide says "Captioning should run on at least one if not all screens showing slides, videos, or live feed of the presenters." If the slides fill the main screen, put the captions on a second screen or TV beside it.

Windows 11, Macs with Apple silicon and PowerPoint have captions built in; live captions on a projector or TV compares them, and PowerPoint live translation covers subtitles inside a slide show.

AI captions on attendees' phones

A screen at the front doesn't help someone in the back row, behind a pillar, or in a room where the slides take the screen. In Live Captions, press Share to phones after Start captions. Show the QR code on the screen, or read out the room code for people to type at www.blink-brief.com/live-captions/listen. An attendee who picks the language being spoken reads the captions as spoken, untranslated, in large type they can resize with A− and A+; one who picks another language reads each line translated a moment after it's said. No app, no sign-up, and it's free during the beta.

  • Limits: up to 200 phones and 3 hours a room, 8 languages at a time, 100 rooms a day per network.
  • One spoken language per room. To switch, press End sharing, pick the new spoken language, then press Share to phones: a new room with a new QR code.
  • Machine translation: Cloudflare's AI translates each line, so check wording that matters with a person.
  • Before anything personal: press Pause sharing, and the phones say sharing is paused until you press Resume sharing.
  • Where the text goes: the caption text goes to Blink Brief's server and to Cloudflare's AI for translation; End sharing deletes it, and a room nobody ends is deleted 24 hours after its last line. See the Privacy Policy.

It is AI captioning, not CART: nobody checks the words, so it doesn't take the place of a captioner someone has asked for. For interpretation into other languages at a larger event, see conference interpretation equipment.

The microphone feed decides the quality

AI captions are only as good as the sound they get. Live Captions' own page: "With a clear microphone close to one speaker they are usually easy to follow, while room echo, music and several people talking at once lower the accuracy."

  • Take a feed from the sound desk into the laptop instead of the laptop's own microphone. A CART captioner wants the same feed.
  • One voice at a time. On a panel, each speaker uses their microphone and waits their turn.
  • Questions from the floor: hand the asker a microphone, or have the moderator repeat the question into theirs, or it won't be captioned.
  • Mute the music on the feed during breaks and walk-ins.
  • Test it in the room before the doors open, with the real microphones and the speaker's voice if you can.

Telling attendees captions are available

  • On the registration form. The Mid-Atlantic ADA Center's guide: "Make sure your registration materials and forms ask specifically about your audience's needs and accommodations." Its sample form has a CART checkbox. ADA.gov allows "reasonable advance notice" but not "excessive advance notice requirements", and says "Walk-in" requests "must also be honored to the extent possible."
  • In the program and on signs: say where the captions show, such as "Live captions on the main screen. To read them on your phone, scan the code on the screen." Don't print a QR code for Share to phones: each share makes a new room and a new code, so the program says "the code will be on the screen".
  • From the stage: at the start of each session, say that captions are on, show the QR code, and read out the room code for people who can't scan.
  • Seats: keep seats with a clear view of the caption screen for the people who asked.

For a church service rather than an event, free live captions for church covers the projector and the livestream.

Questions

What is CART captioning?

CART stands for computer-assisted real-time transcription. ADA.gov describes it as "a service similar to court reporting in which a transcriber types what is being said at a meeting or event into a computer that projects the words onto a screen." A person writes every word live, in the room or remotely.

How much does live captioning cost?

A human CART captioner is paid by the hour. Washington's community and technical colleges list their state-contract rates: on-site CART at $149 or $150 an hour, remote at $79 to $135, depending on the firm. AI captioning services cost less: SyncWords charges $30 for one hour of AI captions at a live event, or $27 an hour on a 60-hour yearly plan; Wordly starts at about $1,500 for a 10-hour package. The free Live Captions tool costs nothing, and Share to phones is free during the beta. Prices read on 7 November 2026.

Is AI captioning ADA compliant?

No tool makes an event compliant on its own; what counts is whether communication with the person is effective. ADA.gov says title II entities (state and local government) must give "primary consideration" to the aid the person requests, and title III entities (businesses and nonprofits open to the public; churches are exempt) are "encouraged to consult" the person. New York State's office for deaf and hard of hearing people says that when CART is requested, AI captions "should never be used" in its place. Ask the attendee what works for them.

Sources