CART captioning is Communication Access Realtime Translation: a trained captioner turns live speech into near-verbatim text shown with a short delay. The US Department of Justice lists real-time captioning among the auxiliary aids that covered entities may need to provide.
The "translation" in CART means turning speech into readable text, often through stenographic shorthand. It does not, by itself, mean translation between spoken languages. An English-language CART booking normally produces English text; ask the provider about language and code-switching needs before booking.
That distinction matters because it decides whether CART is the right ask in the first place. This guide covers what the service delivers, how to assess a captioner, what to include in an accommodation request, and how to plan a booking.
Key Takeaways
- CART is live, near-verbatim speech-to-text by a trained human. Specify the spoken and written languages you need when booking.
- The NCRA Certified Realtime Captioner skills test is five minutes at 180 words per minute and 96% accuracy, without editing. That test score is not a guarantee for a live assignment.
- Under the ADA, public entities give primary consideration to the aid a person requests; covered private businesses should consult with the person where possible. Both must provide effective communication, subject to the applicable limits.
- Published request guidance varies by institution. Ask early, check the organizer's process, and do not assume a universal deadline.
- Choose between CART, sign language interpreting, and multilingual captions based on the individual's communication preference and the event language.
What CART captioning actually is
A CART captioner works a stenotype machine, pressing combinations of keys that stand for whole sounds and syllables rather than single letters. Software maps those chords against a personal dictionary and renders text in near real time. The output goes to a laptop screen turned toward the person who requested it, to a projected caption bar, or to a private browser link on their own device.
The service comes in two shapes. On-site CART puts the captioner in the room with a direct audio feed. Remote CART keeps the captioner elsewhere and pipes audio to them over a phone line or a meeting link, with the text streamed back to a URL. Remote delivery can simplify logistics, but it depends on a reliable audio feed; compare actual provider quotes and test the room microphones.
Verbatim is the defining property. A CART captioner aims to capture the speaker's words, including false starts and interruptions, rather than summarize them. That distinguishes it from meaning-for-meaning services such as TypeWell and C-Print, which condense speech into readable prose. The National Deaf Center explains this service distinction.
How CART compares to adjacent services
| Service | What it produces | Language behavior | Delivered by |
|---|---|---|---|
| CART | Near-verbatim live text | Single language in, same language out | Trained realtime captioner |
| TypeWell / C-Print | Meaning-for-meaning live text | Single language in, same language out | Trained transcriber |
| Sign language interpreting | A signed language, delivered live | Spoken language in, signed language out | Qualified interpreter |
| Automatic captions | Machine-generated live text | Varies by tool; some also translate | Software |
| Post-production captions | Timed caption file for recorded media | Source language, optional subtitle tracks | Captioner or software plus review |
CART is also not interpreting. An interpreter converts meaning between two languages under cognitive load that has nothing to do with typing speed; if that is the service you need, how conference interpreting works across spoken languages covers the modes and the credentials involved. The two services are sometimes booked for the same event, for different people in the same room.
Who is qualified to caption in real time
Credential requirements vary by assignment and jurisdiction; one useful benchmark is available. The National Court Reporters Association's Certified Realtime Captioner credential requires a skills test of literary matter dictated at 180 words per minute, transcribed at 96% accuracy over five minutes, with no editing permitted. Holders maintain NCRA membership and earn 3.0 continuing-education units every three years.
Read that accuracy figure as an exam threshold, not a promised accuracy rate for your event. The NCRA test uses a five-minute sample under defined conditions; real sessions differ in audio quality, terminology, speakers, and duration. Ask providers how they measure and correct live errors instead of comparing unsupported headline percentages.
Accuracy in practice depends on things you control more than on the captioner's certification. Unfamiliar proper nouns, industry acronyms, and overlapping crosstalk are what break a realtime dictionary. Sending the captioner a speaker list and a glossary before the event does more for the transcript than any credential comparison will.
When you are shortlisting providers, ask three questions: which realtime credentials the assigned captioner holds, what their policy is when two people talk over each other, and how they handle a term that is not in their dictionary. The answers tell you more than a headline accuracy percentage.
What the ADA requires, and why the same request lands differently
The ADA lists real-time captioning and computer-aided transcription among possible auxiliary aids. The appropriate aid depends on the individual and the communication context, not on a single mandated method.
Under Title II, which covers state and local governments including public universities, a public entity must give primary consideration to the person's requested aid. It can use another equally effective aid, or account for an undue burden or fundamental alteration. Under Title III, a covered private business should consult with the person where possible and still provide effective communication. The DOJ guidance explains both standards.
When requesting CART, describe the communication barrier, why live text would help, and whether alternatives such as automatic captions would work for you. A public university must give primary consideration to that choice; a covered private venue has a different decision standard. Neither setting promises CART in every circumstance.
The separate Title II web and mobile rule has 2027 and 2028 compliance dates for covered public entities. Those dates concern web and app content; they are not lead times or deadlines for requesting live CART. If a recording is published online, assess its accessibility requirements separately.
How to request CART captioning: two illustrative workflows
Both scenarios below are illustrative, written to show the shape of the process rather than to describe named organizations.
Illustrative workflow one: a semester-long university course
A graduate student contacts the campus disability services office before the term starts, follows its accommodation process, and asks for CART across a fixed weekly schedule. The office assigns a provider for the whole term rather than session by session. The student sends the syllabus and reading list forward so the captioner can preload terminology. A recurring schedule can simplify planning, but changes and provider availability still need coordination.
Illustrative workflow two: a one-off external event
An attendee registers for a two-day industry conference and requests CART on the registration form six weeks out. The organizer is a private company, so Title III consultation applies rather than primary consideration. The attendee names CART explicitly, states what would not work instead, and asks for a written answer by a specific date. Two weeks before the event, they confirm the room's audio feed and ask for a test link if the captioner is remote.
Lead time, and what to send
Published examples differ: Clackamas Community College asks for at least three business days for a meeting, while Boston University strongly prefers three weeks for event requests. These are different settings, not a universal range or legal deadline. Contact the specific organizer early; the DOJ says advance-notice requirements must be reasonable and not excessive.
Whoever you book, send the same package in advance:
- Speaker names and titles, spelled the way they should appear in the transcript.
- A glossary of acronyms, product names, and technical terms specific to the session.
- The agenda or slides, even in draft, so the captioner can anticipate topic shifts.
- Audio details: room microphone setup for on-site work, or the exact meeting link and a test slot for remote.
- Transcript expectations, agreed before the session rather than requested after it.
Ask each provider for a written quote covering billable time, minimum booking, preparation, travel or remote setup, transcript delivery, and cancellation terms. Pricing models vary.
Where CART stops: one language, one reader profile
CART is usually booked for live text in the language being spoken. It does not automatically add spoken-language translation, and the service has to fit the reader's communication preference.
For a multilingual audience, ask separately how each language will be served. A person who prefers sign language may want an interpreter, while another person may prefer text. The National Deaf Center advises matching speech-to-text services to the individual and setting.
A single CART feed generally produces one written-language stream. For several target languages, confirm whether the provider offers a separate translation service and obtain a quote for the whole arrangement.
Where a software layer legitimately fits
This is the honest boundary, and it has nothing to do with price. MirrorCaption is automated software rather than a human CART service. Whether it supports effective communication in a particular accommodation setting requires an individual assessment. It is browser-based real-time transcription and translation: it captures speech from a microphone or, on desktop Chrome and Microsoft Edge, from a shared meeting tab, and produces live text in the reader's own language across 50+ selectable languages. It works alongside browser-based calls on Zoom, Microsoft Teams, Google Meet, and Webex without a bot joining the meeting.
Consider it for informal multilingual meetings and conversations where participants can review the output. For formal accommodation requests, legal proceedings, or clinical settings, follow the organization's requirements and the individual's communication needs; automated captions may not be suitable.
See what live multilingual text looks like
Try live multilingual captions in a browser and check current plan terms before use.
Try MirrorCaption FreeFrequently asked questions
Is CART captioning the same as subtitles or closed captions?
Captions can be live or prepared for recorded media. CART is a human live speech-to-text service; closed captions describe a display format and may be produced in several ways.
Does CART captioning translate between languages?
Not by default. CART turns live speech into readable text in the agreed language; an English-language booking typically produces English text. Request spoken-language translation as a separate service.
How far in advance should I request CART captioning?
Ask as early as possible and check the specific organizer's process. Published examples vary from three business days for a college meeting to a preferred three weeks for some university events; these are not universal legal deadlines.
Can an organization give me automatic captions instead of CART?
Possibly, if the alternative provides effective communication for the person and context. Public entities must give primary consideration to the person's requested aid, subject to recognized exceptions; covered private businesses should consult where possible and still communicate effectively.
Do I get a transcript after a CART session?
Ask before booking. Access to a saved file, editing, privacy, and reuse depend on the provider's agreement and the setting.
Is CART the same as an ASL interpreter?
No. CART supplies live written text; an interpreter conveys communication in a signed language. Ask the individual which format works for them in that setting.
The short version
CART is a human live captioning service. The NCRA CRC skills test sets a five-minute exam threshold of 180 words per minute at 96% accuracy, which is not a live-service guarantee. Under the ADA, the organization must assess effective communication, with different decision rules under Titles II and III.
If you need CART, name it explicitly, contact the organizer early, explain your communication needs, and send the glossary. If you also need translation, say so separately and confirm what the provider can deliver.
For informal multilingual use, try MirrorCaption and see whether live translated text covers what you need. Plan details and current credit allowances are on the pricing page.