Video Conferencing with Live Captions: Best Tools in 2026

The best video conferencing with live captions in 2026 depends on who reads them. Google Meet, Microsoft Teams and Zoom all include same-language captions in roughly 50 to 88 languages, but translated captions usually need a paid tier or add-on. For multilingual meetings, translation-first platforms give each participant captions in their own language.

That answer hides the part that actually decides your purchase. "Has live captions" is now a checkbox every major platform ticks. The real differences sit one layer down: how many languages the captions cover, who needs a license, whether a human captioner can plug in, and how accurate the text stays when real people talk over each other on laptop microphones.

The audience is also bigger than many teams assume. About 15% of American adults (37.5 million) report some trouble hearing, according to the National Institute on Deafness and Other Communication Disorders. Add colleagues who join in their second language, people in noisy open offices and anyone who simply reads faster than they listen, and live captions stop being an edge case.

This guide compares the major platforms on the details that matter, pulls together independent accuracy research, and explains the US accessibility rules that changed in 2026. If you need a searchable written record after the call rather than live text during it, our AI meeting transcription guide covers that side.

Table of Contents

• Live Captions at a Glance: 2026 Comparison
• Three Kinds of Live Captions
• Platform-by-Platform: What Each Tool Offers
• How Accurate Are Live Captions, Really?
• Live Captions for Deaf and Hard-of-Hearing Participants
• Seven Things to Check Before You Choose
• From Live Captions to Full Voice Translation
• Which Live Captioning Setup Fits Your Meetings?
• Frequently Asked Questions

Live Captions at a Glance: 2026 Comparison

The table below summarizes each vendor's own help documentation as checked in September 2026. Language lists change several times a year, so confirm the exact languages you need before you buy.

Two patterns stand out. First, same-language captions are now free or bundled almost everywhere, so they rarely justify a switch on their own. Second, translated captions sit behind a paywall on every big suite, and the license rules differ in ways that change your bill. On Teams, an organizer with Teams Premium or Copilot unlocks translated captions for every participant. On Zoom, the host's account or add-on is what counts.

Three Kinds of Live Captions

The word "captions" covers three different technologies. Mixing them up is the most common reason teams buy the wrong tool.

Automatic same-language captions use speech recognition to turn speech into text in the language being spoken. An English speaker produces English captions. This is what most people mean by live captions, and every platform in the table offers it.

Translated captions add machine translation on top. The speaker talks in German, and you read English. Each participant can often pick a different target language. Quality depends on two error sources stacked together: the speech recognition and the translation.

Human CART captions come from a trained captioner, usually a stenographer, who types what they hear in real time. CART stands for Communication Access Realtime Translation. The US Equal Employment Opportunity Commission lists CART among the accommodations employers may need to provide, and its guidance on hearing disabilities in the workplace includes an example of an employee who requests remote CART for video-conferencing meetings with international clients.

There is also a fourth, informal layer: captions at the device level. Operating systems and browsers can caption any audio on your screen, whatever meeting app plays it. That helps individuals, but the host cannot rely on it, because each participant has to switch it on and supported languages vary by device.

Platform-by-Platform: What Each Tool Offers

Google Meet

Google Meet offers the widest same-language coverage in this comparison. Its help center lists 88 languages and variants that all users can caption, although 62 of them still carry a beta label. You can change the caption font, size, color and background, and you can scroll back through earlier captions. That history only covers the part of the meeting where you personally had captions switched on.

Translated captions cover 69 languages and require Business Standard, Business Plus, Enterprise Standard, Enterprise Plus or Google AI Pro for Education. Google notes the feature "is gradually rolling out and may not be available yet," so test it on your own account before promising it to attendees. Meet Help does not describe a way to route a human CART captioner into the caption track, so accessibility-critical meetings usually need a workaround.

Microsoft Teams

Teams gives live captions to all users in 57 spoken languages and variants. Desktop users can adjust font size and color, show more lines and move the caption position. Teams also supports a human CART captioner, and its sign language view keeps designated interpreters in a fixed, larger video tile.

Translated captions reach 31 target languages with a Teams Premium or Microsoft 365 Copilot license. The license rule is generous: if the organizer holds one, every participant can use translated captions. Privacy teams should read Microsoft's caption FAQ closely. It says the speech recognition service may use the meeting subject, invitation, participant names, attachments and recent emails to improve accuracy. It also states that caption data is deleted after the meeting and that Teams does not save captions unless someone starts a transcript.

Zoom Workplace

Zoom's automated captions cover about 50 languages and regional variants and are enabled by default for paid accounts. Admins can restrict who may start captions, and participants can ask the host to turn them on. For human captioning, the host can type captions, assign a participant to type them, or connect a third-party captioning service with a caption API token.

Translated captions support 39 fully supported languages and need a Business Plus or Enterprise plan, or the Translated Captions add-on. Zoom's documentation flags two limits worth knowing: it does not translate between dialects of one language, such as French from France and Canada, and it warns that automated captions may not be accurate in healthcare scenarios.

Webex

Webex captions and transcribes 16 spoken languages, but free accounts get only five. With the Real-time Translation add-on, captions extend to more than 120 languages. One meeting or webinar can show at most 15 unique caption languages at the same time, which matters for large international events.

Webex has the most explicit CART integration of the big four. Captioning software can post text straight to a meeting's CART API URL, and hosts can assign a participant as the closed captioner. The catch is setup: only the host on a desktop app can copy the caption link for the third-party software, so plan it before the meeting rather than during it.

Meeyra

Meeyra approaches captions from the translation side. Each participant picks their own language, and everyone else's speech appears as captions translated into it, with the speaker's original sentence shown above the translation. Caption lines update while the speaker is still talking, and participants can also hear the translation as an AI voice. It runs in the browser, so guests join from a link without installing anything. See how it works on the live translation page and the full list of 42+ supported languages.

The design has a clear boundary, and it is better to know it upfront. Captions appear when a speaker's language differs from yours. If everyone in the room speaks English and you need captions purely for accessibility, a same-language caption tool or a human CART captioner is the right fit, and Meeyra does not offer a CART integration.

How Accurate Are Live Captions, Really?

Vendors rarely publish accuracy figures for real meetings, and the numbers they do share are hard to compare. Independent research fills part of that gap. Speech recognition accuracy is usually measured as word error rate (WER): the share of words the system gets wrong, drops or invents. A 10% WER means roughly one error every ten words, or one to two mistakes in a typical sentence.

Four findings from this research change how you should set up meetings.

Live is harder than recorded. Researchers at Stuttgart Media University tested 11 speech recognition services on lecture recordings. Streaming transcription, the kind live captions use, averaged a 10.9% WER on English versus 9.37% for batch processing of the same audio. Error rates for individual recordings ranged from 0% to 53.8%. The authors conclude that common services "lack reliability in accuracy." You can read the full study on arXiv.

The microphone matters more than the platform. In OpenAI's Whisper paper, the same model made 16.9% errors on meeting recordings captured with individual headsets and 36.4% on the same meetings recorded by a single distant microphone. That gap is larger than most differences between vendors. A conference room with one ceiling microphone can roughly double the error rate.

Accents and dialects widen the gap. A study in PNAS tested speech recognition from Amazon, Apple, Google, IBM and Microsoft. The systems misrecognized 35% of words from Black speakers and 19% from white speakers. Non-native speakers showed a similar penalty in the Stuttgart study. These tests predate the newest models, but the 2023 Stuttgart tests still found higher error rates for non-native English speakers.

WER is not the whole story for deaf readers. In a user study with 30 deaf and hard-of-hearing participants, Kafle and Huenerfauth showed that a metric weighting important words predicted caption usability better than plain WER. One wrong word in a name, number or negation can matter more than five wrong filler words.

For context, human captioning has its own benchmarks. The National Court Reporters Association's entry-level Certified Realtime Captioner exam requires 96% accuracy at 180 words per minute. UK broadcast regulator Ofcom treated 98% as a good accuracy rate in its live subtitling reviews. Its 2024 access-services guidelines ask broadcasters to aim for an average delay of no more than 4.5 seconds.

Live Captions for Deaf and Hard-of-Hearing Participants

Automatic live captions help many people, but they do not automatically make a meeting accessible. Three rules shape what hosts in the US should expect.

First, the Web Content Accessibility Guidelines require captions for live audio in synchronized media under success criterion 1.2.4. The W3C's explanation of that criterion adds an important nuance. It is not meant to require that every two-way call be captioned regardless of need, and responsibility falls on the callers or the host, not the application.

Second, public bodies have more time than planned. The Department of Justice adopted WCAG 2.1 Level AA as the standard for state and local government websites and apps in 2024. An interim final rule published on April 20, 2026 moved the deadline to April 26, 2027 for entities serving 50,000 people or more and to April 26, 2028 for smaller ones. The Department of Health and Human Services followed with a similar one-year extension for Section 504 recipients: May 11, 2027 for organizations with 15 or more employees.

Third, employers owe individual accommodations. The EEOC guidance says the employee's preference should get primary consideration when several accommodations would work. It also gives an example in which a coworker's notes or a meeting summary do not count as effective, because they do not let a deaf employee ask questions and join the discussion in real time.

In practice, a good accessible meeting combines tools rather than trusting one feature:

• Ask attendees about access needs when you send the invitation, not five minutes before the call.
• Book a human CART captioner or sign language interpreter for high-stakes meetings such as HR conversations, legal matters, medical appointments and performance reviews.
• Share the agenda, names and technical terms with the captioner in advance.
• Ask participants to use headsets and to speak one at a time.
• Pin or spotlight interpreters so their video stays visible during screen sharing.
• Keep the chat open as a backup channel for questions.
• Send a corrected transcript or written summary after the meeting, as a supplement rather than a replacement.

Seven Things to Check Before You Choose

Feature lists make every platform look similar. These seven checks separate them.

1. Your actual languages. Check both the spoken language list and the caption target list. They are often different lengths.
2. Who needs a license. Organizer-level licensing is much cheaper than licensing every participant.
3. Human captioner support. If you ever need CART, confirm the platform accepts it, and whether it disables automatic captions.
4. Customization. Font size, color, background and position matter for low-vision and dyslexic readers, not only deaf users.
5. Caption history. Scrollback lets latecomers and slow readers catch up without interrupting.
6. Data handling. Ask where audio is processed, what context the recognition engine reads, and whether captions are stored.
7. Parity across devices. Browser, desktop and mobile apps often support different caption features. Test the one your attendees actually use.

From Live Captions to Full Voice Translation

Captions solve one problem well: turning speech into text. In a meeting where everyone shares a language, that is often enough. In an international team, same-language captions only move the problem. A Japanese colleague reading fast English captions still has to process a second language at speaking speed, while also watching slides and preparing a reply.

Translated captions go one step further, since each person reads in their own language. The limitation is attention. Reading and watching at the same time is tiring over an hour-long call, and it is hard to hold eye contact with someone while you read their words.

Voice translation closes that gap. You hear the other person in your language and can still glance at captions to check a number or a name. That is the setup Meeyra is built around: captions and an optional AI voice for every participant, in their chosen language, directly in the browser. If you are comparing options, our guides on how to translate a video call and whether Zoom has live translation cover the practical differences. For recurring cross-border meetings, the multilingual meeting guide explains how to run them.

Which Live Captioning Setup Fits Your Meetings?

There is no single best platform, but there is usually a clear best setup for each type of meeting.

For education, captions double as notes for every student, not only those with hearing loss. Our BigBlueButton alternatives guide compares classroom platforms, and the education page shows how Meeyra supports multilingual classes.

The most reliable choice is often a combination: your existing suite for same-language captions, a human captioner when accuracy is non-negotiable, and a translation-first platform when the problem is language rather than hearing. To see translated live captions and voice translation in your own meetings, compare the Meeyra plans or create a free account and use the 60 free translation minutes in a real call.

Frequently Asked Questions

Which video conferencing platforms have free live captions?

Google Meet, Microsoft Teams and Zoom all include same-language live captions without an extra add-on. Webex free accounts caption five spoken languages. Translated captions, where you read a different language than the speaker uses, require a paid tier or add-on on all of the major suites.

Are automatic live captions accurate enough for deaf participants?

Automatic captions work well for clear, one-at-a-time speech on good microphones, but independent tests show error rates above 10% for live, accented or room-microphone audio. For legal, medical, HR or accommodation meetings, a human CART captioner remains the safer choice.

What is CART captioning and when do you need it?

CART (Communication Access Realtime Translation) is live captioning produced by a trained human captioner, usually a stenographer. You need it when accuracy is critical or when a deaf or hard-of-hearing participant requests it as an accommodation.

Do live captions make my meetings ADA compliant?

Not by themselves. Accessibility depends on the participant's needs, and the EEOC expects employers to give primary consideration to the employee's preferred accommodation, which may be CART or a sign language interpreter rather than automatic captions.

What is the difference between live captions and translated captions?

Live captions show speech as text in the language being spoken. Translated captions add machine translation, so each participant can read the conversation in their own language.

Can every participant see captions in a different language?

Yes, on platforms with translated captions. Teams, Zoom, Google Meet and Webex allow this on eligible plans, and Meeyra lets each participant pick their own caption and voice language from 42+ options.

Are live captions saved after the meeting?

Usually not by default. Teams states that it does not save captions and deletes caption data after the meeting, so you need to start transcription separately if you want a record.

Do live captions work in the browser without installing an app?

Yes, on browser-based platforms. Google Meet and Meeyra run fully in the browser. Zoom and Teams also offer web versions, but some caption settings, such as Teams caption styling, are documented for the desktop app.