Localization Accessibility: WCAG, SDH and Audio Description

Accessibility is a translation problem before it is a technical one. A blind viewer needs the visual world described, a deaf viewer needs the dialogue and the meaningful sound in text, and both of them need that in their own language. When a product, a course or a streaming catalogue is localized, the accessibility layer has to be localized with it, and that layer has rules: WCAG sets the standard, SDH subtitles and audio description are the deliverables, and legislation such as the European Accessibility Act and the Americans with Disabilities Act decide when this stops being optional.

This guide explains how accessibility standards apply to multilingual content, what SDH and audio description actually contain, what the law requires (naming the sources), how the work of a translator or localizer changes because of it, and a practical checklist you can run against a real project.

---

1. Why Accessibility Is a Localization Problem

Consider a 40-minute corporate training video. It is produced in English with burned-in on-screen text, an on-camera presenter and a voice-over. It is then localized into six languages. If the accessibility layer is treated as an afterthought, here is what happens: the captions are machine-produced in English only, the on-screen text is never translated into a subtitle file, the audio description is never written, and viewers who cannot hear or cannot see the video get a different, poorer product in every language.

Accessibility is not a feature that sits on top of localization. It is a set of deliverables, and each one has to be produced, timed, translated and quality-checked like any other asset. For the language professional, that is the opportunity: the accessibility layer multiplies the amount of work a single piece of content generates, and it is work that needs a translator, not a captioning robot.

The one line that explains the whole field: accessibility asks who is being excluded, and localization asks which language they are excluded in. The deliverables live where the two questions meet.

---

2. WCAG in Plain Language

WCAG stands for the Web Content Accessibility Guidelines, the international standard published by the World Wide Web Consortium (W3C). It exists as three stable versions: WCAG 2.0, published on 11 December 2008; WCAG 2.1, published on 5 June 2018; and WCAG 2.2, published on 5 October 2023. The versions are backwards compatible, which means content that conforms to WCAG 2.2 also conforms to 2.1 and 2.0. The W3C encourages authors to use the latest version.

WCAG 2.2 is also an approved International Organization for Standardization standard, ISO/IEC 40500:2025, and it is the technical basis most organizations use to meet the European Accessibility Act, together with the European Standard EN 301 549. That is the practical reason a language team should care about WCAG: the law does not name a captioning format, it names a standard, and WCAG is the standard that flows into procurement contracts.

2.1. The four principles

WCAG organizes its guidelines under four principles, usually remembered by the acronym POUR: perceivable, operable, understandable and robust. Each principle holds guidelines, and each guideline holds testable success criteria. The principles are stable and the wording is deliberately plain:

• Perceivable: provide text alternatives for non-text content, and provide captions and other alternatives for multimedia.
• Operable: make all functionality available from a keyboard, and give users enough time to read and use content.
• Understandable: make text readable and understandable, and make content appear and operate in predictable ways.
• Robust: maximize compatibility with current and future user tools, including assistive technology.

For a localization team, two of these principles do most of the work. Perceivable is where captions and audio description live, and Understandable is where reading speed and plain language decide whether the localized version is actually usable.

2.2. Conformance levels

Each success criterion carries a conformance level: A, AA or AAA. Level A is the minimum, AA is the level most public-sector and corporate policies require, and AAA is the enhanced level that is rarely demanded in full. When a client says "we need to be WCAG AA", they are describing a technical target that your subtitles, transcripts and audio description scripts contribute to, even though the full conformance judgment is made on the whole page or product rather than on one file.

---

3. The Success Criteria That Touch Subtitling

WCAG handles multimedia under Guideline 1.2, "Time-based Media", whose intent is stated as providing alternatives for time-based media: audio-only, video-only, audio-video, and audio or video combined with interaction, whether live or prerecorded. The success criterion names themselves tell you which media type they apply to, so a captioning project and an audio-only podcast project read the same guideline and pull different criteria from it.

The criteria that matter most to a language professional are the ones about captions, audio description and alternatives:

Two details in that table are worth pausing on. First, the "alternative for time-based media" is a text document that carries the dialogue and the visual information in order, and it is the Level A route when audio description cannot be produced. Second, extended audio description is the technique where the video is frozen periodically so that more narration can fit than the natural pauses allow. Both are deliverables someone has to write, which is where the localization work begins.

WCAG tells you what has to exist. It does not tell you how to time a cue or where to break a line. Those decisions live in the subtitling standards and platform style guides, such as the public Netflix Timed Text Style Guide and the Amazon Prime Video subtitling guidelines, and in a separate article we compare the rules Netflix, Amazon and Disney actually enforce.

---

4. SDH Subtitles

SDH stands for Subtitles for the Deaf and Hard of Hearing. An SDH file is a subtitle file plus the information a hearing viewer takes for granted: who is speaking, what meaningful sounds occur, and how those sounds and voices are delivered. It is used both intralingually, where the subtitles transcribe the original-language audio, and interlingually, where they translate it for a deaf audience in another language.

4.1. What SDH adds to a subtitle file

• Speaker identification when the speaker cannot be identified visually, using a label or a dash convention.
• Sound effects that carry meaning, written inside square brackets, for example [door slams] or [phone ringing].
• Music where it matters, either as a description ([tense music]) or as lyrics.
• Delivery and tone when it changes the meaning of a line: [whispering], [sarcastic], [laughing].
• Non-verbal communication that a hearing viewer absorbs from the audio, such as laughter, sighs or hesitation.

The editorial test is always the same: does this sound change the viewer's understanding? If it does, it belongs in the file. If it is background texture, it usually does not.

4.2. Reading rules are not optional

SDH has to obey the same timing and reading constraints as any other subtitle file, and the platform style guides state them in numbers. Netflix's general Timed Text Style Guide sets a minimum subtitle duration of five-sixths of a second per event (20 frames at 24 frames per second) and a maximum of 7 seconds per event, using a maximum of two lines. Its language-specific guides add the character limits, and the U.S. English guide sets 42 characters per line. Amazon makes the reading-speed expectations explicit in its Prime Video subtitling guidelines, and we break those numbers down in the companion article.

For an SDH author, the consequence is a real tension: you are adding sound descriptions and speaker labels while respecting a character limit and a reading speed, so condensation decisions become sharper. You can write less dialogue, but you cannot drop a sound cue that changes the plot.

The conventions in this section follow the teaching material of the TranslaStars Closed Captions & SDH course, whose module outline runs from the foundations of captioning and audience awareness to speaker labelling, diversity in speech, paralinguistic features, music and lyrics, sound effects, silence and condensation. The full audiovisual track, from subtitling and dubbing to SDH and voice-over, is collected on the Audiovisual Translation courses page.

---

5. Audio Description

Audio description narrates the visual information that a blind or low-vision viewer cannot see: who enters, what they do, where the scene is, and which on-screen text carries meaning. It is delivered as a separate audio track, sometimes recorded by a voice talent, sometimes synthesized, and it is written to fit the pauses in the original audio.

5.1. The constraints an audio description writer works inside

• It fits the gaps. The narration has to land inside pauses in the dialogue. Where the pauses are too short, extended audio description freezes the video to create room, which is a production decision as much as a writing one.
• It is selective. You describe what carries the narrative, not every visual detail. A costume that signals a character's status matters; the colour of a chair usually does not.
• It does not repeat the dialogue. If a character says "I am leaving", the description does not announce that they leave.
• It is timed to the frame. The description has to sit exactly over the action it describes, and it has to avoid clashing with dialogue and sound cues.

5.2. Translating audio description

When content is localized, the audio description is localized too, which produces two related jobs. The first is translating an existing audio description script, which is constrained translation: the translated line has to fit the same pause as the original. The second is writing audio description from scratch for a market that never had one, which requires the writer to work from the video, not from the source script, because the visual information is what has to be verbalized.

There is also a language question that WCAG states directly: text alternatives and equivalents should match the human language of the content. Captions and audio descriptions are produced in the target language, which is precisely why they are a localization deliverable rather than a post-production afterthought.

---

6. What the Law Actually Requires

Two legal frameworks come up constantly in accessibility conversations, one European and one American. They work differently, and it helps to state each one accurately.

6.1. The European Accessibility Act

The European Accessibility Act is Directive (EU) 2019/882 on the accessibility requirements for products and services. Its purpose is to unify accessibility requirements across the European Union for the products and services judged most important for persons with disabilities, and it covers a defined list that includes computers and operating systems, ATMs and ticketing or check-in machines, smartphones, TV equipment related to digital television services, telephony services and equipment, access to audiovisual media services such as television broadcast and related consumer equipment, passenger transport services by air, bus, rail and waterborne, banking services, e-books and e-commerce.

Two implementation facts are worth keeping straight. Member States had to incorporate the Act into national law by June 2022, and in addressing the Act most organizations use WCAG together with the European Standard EN 301 549, the accessibility requirements for ICT products and services. For a language team, the practical reading is simple: audiovisual media services are named in the scope, and the technical benchmark behind the legal requirement is WCAG.

6.2. The Americans with Disabilities Act and Section 508

The ADA is a federal civil rights law that prohibits discrimination against people with disabilities in everyday activities. It is not a benefit programme and no one applies for coverage. It is organised into five titles, and different titles cover different organisations, which is why the same law reaches a public transport provider in one case and a business open to the public in another.

On the digital side, the Department of Justice has published guidance on web accessibility and the ADA, and it describes the barriers that matter in familiar terms: poor colour contrast, information conveyed by colour alone, missing text alternatives on images, no captions on videos, and inaccessible online forms. For federal agencies, Section 508 of the Rehabilitation Act of 1973 applies: since Congress amended the Act in 1998, federal agencies must make their electronic and information technology accessible, and the standards that came into force in 2018 were harmonised with WCAG 2.0 for web content and ICT.

A caution worth keeping. The ADA and the EAA set requirements for products, services and organisations; they do not hand you a captioning file format. When a client asks for "ADA compliant subtitles" or "EAA ready captions", the accurate answer is to agree the standard and the deliverable, usually WCAG at a stated level plus the platform's own style guide, rather than to treat the law name as a specification.

---

7. How This Changes Your Work

If accessibility is part of the brief, five things change in a typical localization project.

• The asset list grows. You are no longer delivering one subtitle file. You may be delivering subtitles, SDH, a transcript and an audio description script, each with its own timing and its own review.
• Timing is shared. Captions and subtitles often have to share one timing grid, and the audio description has to avoid the same pauses the dialogue uses. One set of time codes is decided first, and everything else follows it.
• The reading constraint bites harder. Adding sound cues and speaker labels consumes characters you do not have, so condensation judgement becomes a core skill rather than an occasional one.
• Conventions become market-specific. Italics for off-screen voices, the format of speaker labels, how sound effects are punctuated: each market and each platform has a house convention, and none of them is the model's default.
• Compliance has to be evidenced. When a client's legal team is involved, the deliverable is judged against a standard, so keeping the project file and the check that was run is part of the job.

---

8. A Practical Accessibility Checklist

Run this list against the deliverable before it leaves your desk.

1. Confirm the target standard and level. Is it WCAG 2.2 at AA, or a platform guide, or both? Write it down before you start.
2. Confirm every required asset. Subtitles, SDH, transcript, audio description, sign-language interpretation: agree which ones are in scope so nothing appears late.
3. Check the timing against the audio. Every cue starts and ends with the speech it represents, and no cue crosses a shot change unless the convention allows it.
4. Check reading speed and duration. Apply the platform's ceiling and the minimum and maximum duration, and flag every cue that breaks them.
5. Check line treatment. Two lines maximum, breaks at syntactic boundaries, and no line that separates an article from its noun or a first name from a last name.
6. Check SDH content. Speaker identification is present where it is needed, meaningful sounds are described, and irrelevant texture is left out.
7. Check italics and conventions. Off-screen voices, on-screen text and song lyrics follow the house style for that market.
8. Check the audio description. It fits the pauses, it does not repeat the dialogue, and it describes what carries the story.
9. Check the language. Captions and equivalents are in the target language of the content, in the correct locale variant.
10. Keep the evidence. Save the project file, the timing grid and the QC note. Compliance conversations always come back to the file.

---

9. Who Hires Freelance SDH and Accessibility Specialists?

The companies that stream accessible content, Netflix, Amazon and Disney among them, do not hire individual SDH subtitlers or audio describers directly. They contract the work to localization vendors, and those vendors recruit and test the freelancers who produce the files. The same is true of broadcasters and of the accessibility teams inside large organizations.

That structure is visible in each company's own documentation. Netflix keeps its timed text and audio description guides in a public Partner Help Center written for its partners and vendors. Amazon's Prime Video subtitling guidelines address their readers as "partners, including subtitling vendors, translators, and quality control specialists". Disney's Subtitle and Closed Captioning Style Guide is distributed to approved vendors and is not published openly.

For a freelancer, the practical consequences are:

• No platform holds a public freelance register. We found no page on which Netflix, Amazon Prime Video or Disney accepts freelance SDH or audio description applications directly.
• Vendors are the entry point. Localization and accessibility vendors hold the accounts, run the tests, and hire the freelancers. Their careers pages and the language-industry job boards are where the roles appear.
• SDH is a qualification, not just a task. Because the conventions in this article are checkable against a style guide, a vendor can test them with a single file, which is good news for a specialist who has trained on them.
• The jobs board is a starting point. Accessibility and captioning roles, including vendor-side ones, are posted on the TranslaStars jobs board.

If a site promises to make you an official subtitler for a streaming platform for a fee, treat it with suspicion: none of the three companies publishes such a route.

---

10. Where to Learn This

The accessibility layer is one of the few parts of subtitling where demand keeps rising and the supply of trained people stays thin, because it combines two skills that are usually taught separately: subtitling craft and accessibility convention.

At TranslaStars, the Closed Captions & SDH course covers the conventions in this article with practical exercises, and the audiovisual programmes are collected on the Audiovisual Translation courses page, which gathers subtitling, dubbing, voice-over, SDH and transcreation in one place. If you want the full route from subtitler to specialist, the Master in AI, Translation and New Tech for Audiovisual Translation combines the craft with the technology. The full catalogue is on the TranslaStars courses page, and accessibility and captioning roles are posted on the TranslaStars jobs board.

Related reading on this blog: Why Audiovisual Translation Is Essential for the market context, Main Subtitling Challenges and Solutions for the everyday obstacles, and the French-language guide Formation au sous-titrage SDH for the SDH training angle.

---

11. Sources and Public Style Guides

These are the public sources behind this article, checked on 5 October 2026.

11.1. Standards and legislation

• World Wide Web Consortium (W3C), Web Content Accessibility Guidelines (WCAG) 2.2.
• Directive (EU) 2019/882 (European Accessibility Act), EUR-Lex.
• Introduction to the ADA and the ADA web accessibility guidance, ada.gov.
• Section 508 of the Rehabilitation Act, section508.gov.

11.2. Platform style guides (public)

• Netflix, Timed Text Style Guide: General Requirements.
• Amazon Prime Video, Subtitling Guidelines.
• Amazon Prime Video, timed text delivery documentation.

Disney's Subtitle and Closed Captioning Style Guide is not public; it is supplied to approved vendors, so it is deliberately not linked here.

11.3. Courses used as a source

The SDH and accessibility conventions described above follow the TranslaStars programme outlines:

• Closed Captions & SDH: Essential Skills for Translators & Subtitlers (Jessica Hinds-Mingo, 10 hours), whose modules cover the foundations of captioning, audience awareness, speaker labelling, diversity in speech, paralinguistic features, music and lyrics, sound effects, silence and condensation.
• Audiovisual Translation Foundations: Subtitling & Dubbing for the subtitling and dubbing basics that support SDH work.
• Master in AI, Translation and New Tech for Audiovisual Translation for the audiovisual, accessibility and technology track.

The full audiovisual catalogue is collected on the Audiovisual Translation courses page.

---

Frequently Asked Questions

11.1. Is SDH the same as closed captions?

In practice they describe the same kind of file: dialogue plus speaker identification and sound information. The historical difference is that captions were developed for same-language broadcast in the United States, while SDH is the term used internationally and especially when the file is translated for deaf and hard-of-hearing viewers in another language. For a working professional, treat them as one skill with two contexts.

11.2. Does accessibility apply to translated content, or only to the original?

It applies to the localized product as well. WCAG states that text alternatives and equivalents should match the human language of the content, so captions and audio descriptions are produced in the target language. A localized video without localized captions is not accessible to the audience it was localized for.

11.3. Do I need to be a translator to write audio description?

You need both skills. Writing audio description from scratch is close to scriptwriting, because you decide what visual information carries the narrative and phrase it to fit the pauses. Translating an existing audio description script is constrained translation, because the translated line has to fit the same window as the original. Most professionals arrive from subtitling and learn the accessibility conventions on top.

11.4. What is the difference between the EAA and WCAG?

The European Accessibility Act is legislation: it sets out which products and services have to be accessible and requires Member States to enforce that. WCAG is the technical standard that describes how to build accessible content. In practice, organizations meet the Act by applying WCAG and the European Standard EN 301 549. One is the requirement, the other is the instruction manual.

11.5. Where do I start if I already subtitle?

Start with the SDH conventions, because they are the closest extension of what you already do: speaker identification, sound description and italics sit inside the timing and reading rules you already apply. Then learn audio description, which is a writing skill you can practise on any short video with a transcript and a stopwatch.