Live Captions for Video Calls: The Right App for Each Need

Live captions for video calls come from three places: the call app, your device, or a human captioner. iPhone, Android, Windows, Mac and Chrome all include free, on-device live captions that work in most call apps. The right choice depends on why you need them: hearing, noise or language.

The demand is broader than most people assume. The World Health Organization estimates that 430 million people need rehabilitation for disabling hearing loss, a figure projected to pass 700 million by 2050. Yet in a May 2026 survey of 3,000 viewers in the US, UK, France, Germany and Spain, 81% reported no hearing difficulty while 80% used subtitles at least sometimes. Video calls are where that habit meets live, unscripted speech, and that is where caption quality starts to matter.

This guide maps each of those needs to the caption source that actually meets it. You will see what the five live captions apps already built into your devices can and cannot do in September 2026, when automatic captions stop being good enough, and where translated captions take over once people no longer share a language.

Table of Contents

• Three Reasons People Turn On Live Captions
• Where Live Captions Come From
• Live Captions Apps Already on Your Devices
• When Automatic Captions Are Not Enough
• Captions Across Languages: Where Translation Fits
• Make Any Call Easier to Caption
• Privacy: On-Device vs. Cloud Captions
• Frequently Asked Questions
• Which Live Captions Setup Fits Your Call

Three Reasons People Turn On Live Captions

Every caption stream looks the same on screen: white text under a video tile. The people reading it want very different things, and that difference decides which tool you should use.

Hearing Loss: Captions as Access

For someone who is deaf or hard of hearing, captions are not a convenience. They are the call. The WHO counts more than 25% of people over 60 as affected by disabling hearing loss, and the US National Institute on Deafness and Other Communication Disorders reports that about 15% of American adults (37.5 million) have some trouble hearing.

This group sets the highest quality bar. The UK regulator Ofcom used 98% accuracy as its quality threshold for live TV subtitles and recommends a delay of no more than three seconds. Realtime captioners certified by the NCRA must reach 96% accuracy at 180 words per minute, with no editing allowed. Speaker names and sound cues such as laughter matter too, because a caption without a speaker is a sentence without context.

Noise and Quiet: Captions as Backup

The second group hears fine but cannot use sound at that moment: a commuter on a train, a parent next to a sleeping child, someone in an open-plan office, or anyone stuck with a colleague whose microphone sounds like a tin can. In a 2023 YouGov survey, 33% of American subtitle users said they switch subtitles on in noisy environments, 22% to avoid disturbing others and 40% to understand accents. Home offices are not quiet either: in a study of 1,934 remote workers in Italy, 57% described themselves as sensitive to noise.

The bar here is lower. A caption that gets the gist right and appears without setup is enough, because the reader can usually hear part of the audio or ask for a repeat. Device captions suit this need best, since you can switch them on in seconds without asking anyone.

Language: Captions as a Second Channel

The third group follows a call in a language they speak, but not natively. For them, captions work as a second channel: reading supports listening, especially with fast speakers, strong accents and unfamiliar names. A 2013 meta-analysis of 18 studies found a large effect of captions on second-language listening comprehension, and a review of more than 100 studies concluded that captions particularly benefit people watching in a non-native language.

Captions only help if they are accurate and fast enough, though. Research on non-native speakers found that automatic transcripts with a 10% word error rate improved comprehension, while transcripts above 20% brought no improvement. In the same line of work, transcripts delayed by four seconds did not help at all. And when participants share no language, same-language captions cannot help; you need translated captions instead.

Where Live Captions Come From

Before you compare apps, separate the four places a caption can come from. Each one is controlled by a different person.

1. The call app. Most meeting platforms generate captions in the cloud. The host, the admin or the paid plan decides whether they are available and in which languages. Our comparison of video conferencing platforms with live captions covers those plan and language differences in detail.
2. Your own device. The operating system or browser listens to whatever audio your device plays and captions it locally. You switch it on yourself, and it works across apps.
3. A human captioner. Communication Access Realtime Translation (CART) providers type a verbatim transcript live, usually streamed through a separate link or a caption integration. It costs money and needs booking, but it holds up where software struggles.
4. A translation layer. Speech is recognized, translated and shown as text in another language. This is the only source that helps when two people have no language in common.

The listener-side option is the one most people overlook. You do not need the host's permission, a specific plan or even the same app as the other person.

Live Captions Apps Already on Your Devices

All five major platforms now ship their own live captions. We checked each vendor's official documentation in September 2026, after iOS 27 and macOS 27 (Golden Gate) were released on September 14.

iPhone, iPad and Mac

Apple's Live Captions transcribe audio from any app, so they work in third-party call apps as well as in FaceTime and the Phone app. In FaceTime, the caption window also shows who is speaking. Apple supports 18 locales in iOS 27, including German, French, Spanish, Japanese and several English variants, but not Turkish. On a Mac, Live Captions caption all apps by default, and Type to Speak reads out what you type during a call.

Live Translation is a separate Apple Intelligence feature. It translates one-on-one FaceTime and phone calls between nine languages, German included, on iPhone 15 Pro or later. Turkish is supported for Live Translation in Messages, but not for calls.

Android Phones

Android Live Caption covers 15 languages on Pixel 6 and later and on selected phones running Android 14 or newer, including German and Turkish. It captions phone calls, video calls and voice messages, and you can type replies during a call on Pixel 6 or later. Google adds two limits: it "isn't intended for calls with more than one other person", and some calling apps block it.

Expressive Captions, currently English only, add sound labels such as laughter or applause and capitalize shouted words. For translation, Pixel 10 and later translate calls between English and ten languages, without Turkish. Samsung's Galaxy AI Live Translate supports both German and Turkish in the Samsung dialer and some third-party apps.

Windows 11

Windows Live Captions caption any audio that passes through your PC, which makes them the most app-agnostic option on a laptop. Recognition covers 21 locales, including German but not Turkish. Two practical details matter in calls: microphone captions are off by default, and captions follow your default output device, so set your headset as the default before the call starts.

On Copilot+ PCs, Live Captions also translate more than 40 spoken languages into English, Turkish and German included. Your Turkish-speaking client therefore appears as English captions, even though Windows cannot caption Turkish as Turkish.

Chrome

Chrome's Live Caption captions audio playing in the browser, and Google's enterprise documentation says it works with "most web-based video or audio chat services". Speech recognition runs locally. Google publishes a language list for Chromebook Plus that includes German and Turkish; on Windows and Mac you add caption languages in Chrome's settings. The optional Live Translate step is different: it sends caption text to Google for translation.

When Automatic Captions Are Not Enough

Automatic captions are good, but not uniformly good. An independent benchmark of 11 speech recognition services on university lectures, published in ACM Transactions on Accessible Computing, measured an average word error rate of 7.0% for native English speakers, 10.2% for speakers of English as a second language and 15.3% for German (arXiv). Live streaming recognition did worse than processing a finished recording: 10.9% versus 9.37%. On a call, roughly one word in ten can be wrong.

Whether that is acceptable depends on who is reading. In focus groups with 25 deaf and hard-of-hearing participants, the same research group found that ratings dropped significantly between a 4.9% and a 14.7% error rate, and that 9.3% reached only medium acceptance (arXiv). Delay matters as much as accuracy. In a September 2026 Gallaudet University study with 216 deaf and hard-of-hearing participants, synchronized automatic captions scored 5.20 on a 7-point scale, while the same captions with a two-second delay fell to 4.81 (arXiv).

Use the four tests the FCC applies to TV captions in 47 CFR 79.1 as your checklist: captions should be accurate, synchronous, complete and appropriately placed. If a setup fails one of them in a test call, it will fail in the call that matters.

Book a human CART captioner when the stakes are high: job interviews, medical and legal appointments, performance reviews and training with assessment. In the US, the EEOC's guidance on hearing disabilities in the workplace lists remote CART as a possible accommodation. Keep device captions running as a backup, and let the person who relies on the captions choose the setup.

Captions Across Languages: Where Translation Fits

Once people do not share a language, captions need a translation step. That pipeline does more work: speech is recognized, split into sentences, translated and then displayed. Each stage adds a little delay, and recognition errors carry through into the translation. Showing the original sentence next to the translation lets bilingual readers spot those mistakes.

The built-in options stop at the one-to-one call. Apple's Live Translation and Pixel's call translation handle two people and a fixed set of languages, and Windows translates only into English or Simplified Chinese. A group call where a Turkish supplier, a German engineer and a Brazilian buyer each want captions in their own language needs translation at the meeting level.

That is the job Meeyra's live translation is built for. Each participant selects their language, and speech is translated into 42+ languages in real time. Captions show the original sentence above the translation, words appear as they are recognized, and earlier captions stay in a side panel during the call. You can read the captions, listen to a translated voice, or both. Meeyra runs in the browser, and guests join through a link without installing anything.

Meeyra's captions are translation captions, though. When a speaker uses the same language you selected, no caption appears. For a deaf participant who needs same-language captions, pair Meeyra with Chrome Live Caption or Windows Live Captions on the same computer, or book a CART captioner. For the wider workflow, see our guide on how to translate a video call.

Make Any Call Easier to Caption

Caption quality is decided before any software runs. These eight habits raise accuracy for every source, from device captions to a human captioner.

1. Move the microphone close. A headset or a mic within arm's length beats a laptop mic across the room. In the Whisper speech recognition paper, the error rate on the same meeting recordings rose from 16.9% with headset microphones to 36.4% with a single distant microphone (arXiv).
2. Remove noise at the source. Close the window, mute notifications and use noise suppression. Our guide to reducing background noise on your mic walks through each layer.
3. Speak one at a time. Crosstalk merges two voices into one garbled line. In group calls, the host should hand over the floor explicitly.
4. Say your name first in group calls. Device captions rarely label speakers, so "This is Dana" at the start of a turn helps everyone who reads.
5. Share names and jargon in advance. Post client names, product names and acronyms in the chat before the call. Human captioners use this list; software users can check it against the captions.
6. Keep faces visible and well lit. Many people with hearing loss combine captions with lip-reading. Our guide to the best lighting for video calls shows how to light your face evenly.
7. Finish your sentences. Translation engines work sentence by sentence, so a short pause at the end of a thought makes translated captions arrive sooner.
8. Ask, don't assume. Ask the person who relies on captions which setup works for them, and agree on it before important calls.

Privacy: On-Device vs. Cloud Captions

Here the built-in apps have a clear advantage. Apple, Google (for Android Live Caption and Chrome's recognition) and Microsoft all state that audio for their live captions is processed on the device. Google's Android documentation goes furthest: audio and captions "are never stored or sent to Google". The exceptions sit at the edges: Chrome's Live Translate sends caption text to Google, and captions inside meeting platforms are usually generated in the provider's cloud.

Saved transcripts are a different matter from live captions. On iPhone, keeping call captions for an hour triggers a spoken notice to everyone on the call that the call is being transcribed. If you store captions, treat them like a recording and follow the consent rules that apply to you.

One more rule for workplaces: a request for captions can reveal a disability. Under the GDPR that is health data with special protection, and in the US the ADA requires employers to keep medical information from accommodation requests confidential. Handle caption requests discreetly, and never announce why someone has captions switched on.

Frequently Asked Questions

Can I get live captions on any video call app?

Usually, yes. iPhone, Mac and Windows caption audio from any app, and Chrome captions most web-based calls. Android Live Caption works in phone and video calls, but some calling apps block it and it is not designed for group calls.

Which languages do built-in live captions support?

It depends on the device. Apple and Windows caption German but not Turkish, while Android Live Caption and Chrome support both. English, French, Spanish and Japanese are available on all five platforms.

Are live captions on my phone private?

Built-in live captions on iPhone, Android, Mac, Windows and Chrome process speech on the device itself. Privacy changes when you add cloud translation or store transcripts, so check those settings separately.

How accurate are automatic live captions?

Independent tests put automatic captions at about 7% word errors for native English speakers and around 10% for non-native speakers or live streaming. A close microphone, one speaker at a time and a quiet room bring those numbers down noticeably.

Are automatic captions enough for a deaf participant?

For everyday calls, often yes. For job interviews, medical appointments, legal matters or assessments, a human CART captioner is the safer choice, ideally with device captions running as a backup.

What is the difference between live captions and translated captions?

Live captions show what someone says in the language they speak. Translated captions show the same speech in another language, which adds a translation step, a little more delay and a second source of errors.

Does Meeyra caption speakers who talk in my own language?

No. Meeyra's captions are translation captions: they appear when a speaker uses a different language from the one you selected. For same-language captions in a Meeyra call, turn on your device's live captions or book a CART captioner.

Which Live Captions Setup Fits Your Call

Match the setup to the need, not to the app you happen to be using:

• Everyday call and you are hard of hearing: switch on your device's live captions and keep the other person's face visible.
• Interview, medical, legal or graded session: book a CART captioner and keep device captions as a backup.
• Noisy or must-stay-quiet surroundings: device captions, plus a headset if it happens every day.
• A colleague follows in their second language: same-language live captions, a slower pace and a shared list of names and terms.
• No shared language: translated captions that keep the original sentence visible.
• Group call across several languages: meeting-level translation rather than one-to-one call features.

Most of these setups cost nothing, because the live captions app you need is already on your phone or laptop. What device captions cannot do is carry a group across a language barrier. If that is your situation, create a free Meeyra account; new accounts include 60 minutes of AI translation, and the pricing page lists the plans with translated captions.