What Are Captions? A Complete Guide to Types, Uses, and How They're Made (2026)

Captions — the short answer

Expanded definition — what captions actually contain

A caption track for a video is more than a transcript on screen. It reproduces the full audio experience in text so a viewer who can't hear the audio understands what's happening. Concretely, a well-produced caption track includes:

• Spoken dialogue — verbatim or lightly edited for readability, timed to appear as the words are spoken.
• Speaker identifications — LISA:, MARK (V.O.):, NARRATOR: — especially when multiple speakers appear off-screen or unseen.
• Sound effects — [door slams], [phone ringing], [thunder], [laughter]. Anything that adds to plot, mood, or continuity.
• Music cues — [upbeat rock music], [♪ You Are My Sunshine ♪], [ominous instrumental]. Music that sets mood or contains recognizable lyrics.
• Non-speech vocalizations — [sighs], [laughs nervously], [gasps], [groans] — when they convey emotion or meaning.
• Foreign-language dialogue — either translated in the caption itself or bracketed as [speaking Spanish] when the audience isn't meant to understand it.
• Off-screen sound — [distant siren], [car engine approaching] — sounds that affect the scene but aren't on camera.

What captions do NOT typically include: purely ambient audio that doesn't affect the scene, background chatter without narrative purpose, or minor sounds that would clutter the caption pane without serving comprehension. Style guides (Netflix Timed Text, BBC Subtitle Guidelines) explicitly recommend omitting these.

The five types of captions in modern use

Captions exist in five practical forms today, distinguished by how they're delivered and whether the viewer controls their visibility. The right type depends on the platform, the audience, and the delivery constraints. Extraction table:

For the closed-caption-specific deep dive — CEA-608/708 technical detail, how to add or remove CC on TV/YouTube/streaming, the trademark history of the “CC” symbol — see what is closed captioning?

Captions vs subtitles — the practical difference

The single most common confusion around captions is how they differ from subtitles. Both are on-screen text; both accompany video. The difference is the viewer they're designed for:

• Captions assume the viewer cannot hear the audio. They include dialogue plus non-speech information (sound effects, speaker IDs, music cues).
• Subtitles assume the viewer can hear the audio but doesn't understand the spoken language. They include dialogue only.

Regional terminology matters. In the United States, “captions” and “subtitles” are two distinct words for two distinct things. In the UK, Ireland, Australia, and much of continental Europe, “subtitles” is used for both accessibility and translation — BBC subtitles, for instance, include speaker IDs and sound effects (they're accessibility captions in US terminology). Context and delivery platform disambiguate.

For the full comparison — including SDH, closed vs open, US/UK/EU legal frameworks, and how to decide which your video needs — see captions vs subtitles.

Where you see captions — platform by platform

Captions are now a default expectation across nearly every video platform in 2026. The specific implementation varies by platform:

How captions are created — five production methods

The captioning pipeline you pick trades accuracy for cost, speed, and latency. The five common methods in 2026:

For most creator workflows in 2026, AI-only transcription is the pragmatic choice — 90-95% accuracy on clean English audio at pennies per minute, exported as SRT or VTT and imported into any video platform. VexaScribe generates captions using OpenAI's Whisper Large-v3 in 99 languages; see caption generator for the tool page and how accurate is Whisper for the underlying accuracy math. For broadcast, legal, or medical compliance, budget for AI + human review or full human transcription instead.

Who uses captions — well beyond accessibility

Captions were designed for deaf and hard-of-hearing viewers. In 2026 they're used by a much broader audience. A widely-cited Ofcom study (2006) found roughly 80% of UK caption users were not deaf or hard-of-hearing. Modern caption usage rates on Netflix, YouTube, and Amazon Prime run 50-80% among general audiences.

Who actually turns captions on today:

• Deaf and hard-of-hearing viewers — the original and primary audience; captions are essential accessibility.
• Silent-autoplay viewers — Facebook, Instagram, TikTok, and LinkedIn autoplay videos muted. 85%+ of Facebook video is watched with sound off (Digiday, Facebook internal data); captions are required for engagement, not optional.
• Viewers in loud environments — gyms, public transport, open-plan offices, coffee shops, bars, factories.
• Viewers in quiet environments — bedtime, hospitals, libraries, sleeping partner nearby, sleeping baby.
• Non-native English speakers — captions bridge the comprehension gap for accented speech, technical vocabulary, or rapid dialogue.
• Language learners — captions in the target language reinforce vocabulary and pronunciation.
• Viewers with attention or processing differences — captions provide a redundant channel that supports comprehension for many neurodivergent viewers.
• Viewers of accented or technical content — even native speakers catch details more reliably by reading than by listening in rapid or heavily-accented speech.
• Content re-purposers — captions get pulled into transcripts for SEO, blog posts, podcast show notes, and searchable video archives.

For creators and businesses, captions have crossed the threshold from accessibility feature to core distribution feature — video without captions loses reach on social, muted-autoplay platforms, and reduced-accessibility audiences.

Legal requirements at a glance — US, UK, EU

Captioning law varies by jurisdiction, platform, and content type. This is a summary — for a deeper legal breakdown, see captions vs subtitles which covers US, UK, and EU frameworks in detail. Consult a lawyer for specific compliance questions.

The trend line: captioning requirements are expanding, not contracting. The EAA (June 2025) brought EU e-commerce and audiovisual media services under a common accessibility framework. ADA case law in the US continues to extend captioning obligations to online video. WCAG 2.2 (2023) reaffirmed 1.2.2 (captions on prerecorded video, Level A) and 1.2.4 (captions on live video, Level AA) as accepted global baselines.

File formats used for captions

Captions are delivered in different formats depending on the platform. For most creator workflows in 2026, SRT is the right default — supported by YouTube, Vimeo, Premiere, DaVinci, CapCut, VLC, and virtually every platform. Broadcast and OTT delivery use specialized formats.

Practical recommendation for creators: generate an SRT first — it works everywhere. Convert to WebVTT with a tool like our SRT ↔ VTT converter if you're embedding via HTML5 <track>. Use SCC or TTML only when delivering to US broadcast or professional OTT pipelines. For creating an SRT from scratch, see how to create an SRT file.

Captions vs transcripts — related but different

A common conflation. Captions are time-synced text that appears on screen at the moment the audio plays. Transcripts are plain-text records of what was said, without timing or synchronization to video. Both come from the same source material — often produced together — but they serve different jobs.

• Captions serve accessibility during video playback. They're embedded or attached to the video and appear synchronized to speech.
• Transcripts serve searchability, SEO, and use cases where the reader isn't watching the video — podcast show notes, meeting summaries, legal records, LLM inputs.
• Most modern captioning tools produce both from a single transcription pass: word-level timestamps generate SRT captions on export and TXT/DOCX transcripts on separate download.

For the transcription-first perspective (what the audio-to-text step actually does), see what is audio transcription? For plain-text output workflows, see transcribe audio to text.