SDH Subtitles and Audio Description: Accessibility Translation

Accessibility translation is the collective name for the services that close the sensory gaps standing between a viewer and the content. It has two main branches: SDH subtitles for deaf and hard of hearing viewers, and audio description for blind and partially sighted viewers.

Contents

Both sit within the translation industry, but they are produced with different skills. In SDH subtitling the job is to put the information carried by the sound into writing. In audio description the information carried by the picture has to be turned into speech.

Below we look at the rules of both services, their production workflow and the items to watch when budgeting for them.

What are SDH subtitles, and how do they differ from ordinary subtitles?

Ordinary subtitles assume that the viewer can hear the sound. They convey speech and nothing else; they do not write out the music, the sound of a door or who is speaking, because those things are already audible.

SDH subtitles remove that assumption. They put the whole of the information carried by the sound into written form. The expansion of the abbreviation says as much: subtitles for the deaf and hard of hearing.

The practical consequence of the difference is that two separate files are produced for the same content. The SDH version carries more text, and that changes the reading-speed calculation. Once the extra information is added, some blocks have to be split and others shortened.

Another difference concerns which language is involved. SDH subtitles are most often produced in the same language, which makes them a piece of intralingual transfer rather than a translation. With foreign content, the translation layer and the accessibility layer are built together and the workload rises.

What is shown in SDH subtitles?

The scope of SDH subtitling is set by a single question: which piece of sound information carries meaning? Not everything audible is written down; only the elements that bear on the narrative are conveyed.

The first category is speaker identity. In scenes where the speaker is off screen, or where more than one person is talking, the subtitle states who is speaking. A name label, a colour code or positioning is used for this, and which of them is preferred depends on the platform’s style guide.

The second category is ambient sound. Sounds that bear on the plot, such as a doorbell, a phone vibrating or a scream from behind, are written inside square brackets. Background sounds that add nothing to the narrative are left out; otherwise the screen fills with unnecessary text.

The third category is music and tone. The presence and the character of music that carries the mood of a scene are noted, and lyrics are conveyed where they form part of the narrative. If the tone of the speech – a whisper, a shout, a sarcastic edge – changes the meaning, a short note is added as well.

How is audio description produced?

Audio description is an additional narration layer prepared for blind and partially sighted viewers. It conveys the information that appears on screen but is never spoken.

Production begins with the script. The describer watches the content, identifies the gaps free of dialogue and writes narration sentences that will fit into them. The elements to be described are ranked by importance: the setting, the people, the action, the text shown on screen and the visual details that are critical to the narrative.

The second stage is voicing. The narration is usually delivered in a neutral tone, without interpreting the events. The aim is to let viewers draw their own conclusions: describing the facial expression is preferred to saying that someone looks frightened.

The third stage is mixing. The description voice is balanced against the original soundtrack and must not run over the dialogue or over important effects. Delivery can be made as a separate audio track or as a full mix with the description already in place.

Time and language constraints in audio description

The hardest constraint on description is time. The narration can only fit into the gaps free of dialogue, and those gaps are plentiful in some productions and almost non-existent in others.

When the gap is tight the describer has to choose: which information is essential if the narrative is to make sense, and which is decoration? That choice is the core skill of the job and it develops with experience. A wrong choice leaves the viewer building an incomplete picture of the scene.

On the language side, economy is essential. Sentences should be short, verbs in the present tense and words concrete. In Turkish that means avoiding chains of suffixes and breaking up long noun phrases; the agglutinative structure pushes the syllable count up quickly in long sentences and will not fit into narrow gaps.

There is also terminological consistency. The same person, place or object should be called by the same name every time. In series and serialised productions that consistency has to hold across episodes as well, and a description glossary is kept for the purpose.

Production workflow and delivery formats

The workflows of the two services differ and should be budgeted separately. The table below summarises the main differences.

Criterion	SDH subtitles	Audio description
Target viewer	Deaf and hard of hearing viewers	Blind and partially sighted viewers
Output	Time-coded text file	Audio track or description script
Main constraint	Reading speed and character limit	Length of the dialogue-free gaps
Is a studio needed?	No	Yes, for voicing and mixing
Delivery format	SRT, WebVTT, TTML	WAV/MP3 track or mix

On the delivery side the platform’s style guide is decisive. Some platforms list SDH subtitles as a separate language option; others ask for two separate subtitle tracks in the same language. For the description too, whether a separate track or a ready-made mix is expected should be asked at the outset.

Quality control is a step of its own in both services. On the SDH side, reading speed, overlaps and tag consistency are checked automatically. On the description side, whether the narration runs over the dialogue and whether the levels are balanced are checked in a listening pass.

Common mistakes and quality measures

Accessibility content is rarely checked by its target audience once it has been produced. Mistakes therefore go unnoticed for a long time.

The most common mistake on the SDH side is treating automatically generated subtitles as accessible. Automatic text can convey the speech, but it does not produce speaker distinctions, ambient sounds or tone information, and errors in punctuation and proper names distort the meaning as well. The second common mistake is adding more sound information than is needed and filling the screen until it can no longer be read.

The most common mistake on the description side is adding interpretation. Saying that someone is looking on sadly draws the conclusion on the viewer’s behalf; the right approach is to convey what can be observed. The second mistake is narration running over the dialogue, which makes both the description and the dialogue unintelligible.

The most reliable quality measure is feedback from the target audience. A single review pass with a small group of users brings out problems that technical checks cannot catch, and it builds a lasting guide for later productions.

Budgeting and planning

In accessibility services the cost varies with the length and the density of the content. Two videos of the same length can call for very different amounts of work.

Establish the length of the content and the density of the speech; this is what shows whether there is room for description.
If you already have a subtitle file, share it; it gives SDH production a starting point.
Clarify which platform’s guide the work will follow.
Decide how speaker distinctions will be marked: name label, colour or position.
State whether you want voicing for the description or only the script.
Set out the delivery format and the audio track structure in writing.
Leave room in the schedule for quality control and one round of corrections.

Once these seven points are clear, quotes become comparable. Where the scope is left vague, differences in price are misleading; a low quote usually stands for a narrower scope.

The Türkiye context

Accessibility in Türkiye is an active field in terms of both legislation and practice. In broadcasting, the regulation brought into force by the Radio and Television Supreme Council in 2019 set out the procedures and principles for making broadcast services accessible to viewers with hearing and sight impairments. Accessibility criteria are also appearing more and more often as a requirement in the digital content of public institutions.

On the language side, two details specific to Turkish stand out. The first is that the text grows as sound information is added to SDH subtitles; because of the agglutinative structure, descriptive notes can take up more characters than expected and the reading-speed limit is strained quickly. Choosing short, economical wording is a necessity here.

The second is the sign language layer. Turkish Sign Language is a language independent of written Turkish; sign language interpreting does not take the place of subtitles and is planned as a separate service. When the scope of accessibility is being set, these three layers – subtitles, description and sign language – need to be assessed one by one. This article is intended as general information and is not legal advice.

What does this mean for you?

In most organisations accessibility either never comes up at all or is remembered just before the publication date. Both outcomes are costly: in the first, part of the audience cannot reach the content; in the second, the work is rushed.

The practical approach is to make accessibility part of the content production plan. Leaving gaps for description when a video shoot is being planned, producing the subtitle file with time codes from the start and keeping a shared term list all bring later costs down noticeably.

The order to start in is this: first take an inventory of which of your existing content has subtitles, then produce SDH versions for the most watched items, then bring description into the plan for new productions. We can work out together what is needed in terms of scope and delivery format: write to us or look through the related articles in our knowledge centre.

Frequently asked questions

What is the difference between SDH subtitles and ordinary subtitles?

Ordinary subtitles convey speech only and assume that the viewer can hear the sound. SDH subtitles also write out the information carried by the sound: speaker identity, important ambient sounds, music and tone.

What is audio description?

It is a narration layer for blind and partially sighted viewers, placed in the gaps between the dialogue. It conveys the information that appears on screen but is not spoken: the setting, the people, the movement, the written text and the important visual details.

Are these services a legal requirement?

There are regulations in the broadcasting field, and the services can be stipulated in public tenders. For corporate content the answer usually depends on the contract or on an accessibility policy. The scope has to be assessed project by project.

Can we convert our existing subtitles to SDH?

Yes, existing subtitles are a good starting point. Speaker distinctions, sound information and timing are all reworked, though, and because the added text changes the reading speed the blocks are reviewed as well.

Sources

Pietrzak, P. & Kornacki, M. (2021). Using CAT Tools in Freelance Translation. Routledge, pp. 17–18.
Neves, J. (2005). Audiovisual Translation: Subtitling for the Deaf and Hard-of-Hearing. PhD thesis, Roehampton University, pp. 120–160.
Fryer, L. (2016). An Introduction to Audio Description: A Practical Guide. Routledge, pp. 15–60.
W3C (2018). Web Content Accessibility Guidelines (WCAG) 2.1 – 1.2 Time-based media criteria.
Radio and Television Supreme Council (2019). Regulation on improving access to broadcast services.