Video Production
Video subtitles: clear communication when sound is unavailable
Subtitles help when viewers have no sound, the surroundings are noisy or spoken content is hard to follow. Automatic transcription is a starting point, not final approval.
Video subtitles are often generated automatically at the very end of production. That is where avoidable errors appear: names are wrong, technical terms are replaced with similar-sounding words and long sentences disappear before anyone can finish reading them. The video has been published, but the text no longer matches the voice.
Good subtitles are editorial work. They make speech usable in quiet and noisy settings and support people who cannot fully hear or process the spoken content. This does not guarantee twice the reach or any other fixed performance gain. Their practical value does not need an inflated promise.
Updated 29 July 2026
Contents
- Subtitles, captions and text burned into the picture
- Which videos need subtitles?
- Turn a transcript into readable subtitles
- Check timing and presentation
- Separate file or text in the export?
- Translate without losing the meaning
- Use a dependable approval process
- Frequently asked questions
Subtitles, captions and text burned into the picture
Several types of on-screen text are described as subtitles in everyday conversation. Distinguishing them helps a production team ask for the right deliverables.
Subtitles reproduce spoken language as text. Closed captions can also identify speakers and describe relevant sounds. Viewers can switch them on and off when the platform supports a caption track. Open captions are part of the picture and remain visible in every playback.
Graphic titles have a different job. They name a person, show a figure, mark a chapter or give a call to action. A video with large animated keywords does not necessarily provide a complete text version of the speech.
| Format | Benefit | Limitation |
|---|---|---|
| Separate subtitle file | Can be switched on and off, often searchable and easier to correct | The platform must support the file and format |
| Burned-in captions | Visible in every copy of that export | Cannot be disabled and corrections require a new export |
| Automatic platform captions | Available quickly | Wording, punctuation and timing still need checking |
| Manually prepared captions | Speakers and useful sound information can be identified accurately | Needs more editorial time |
The right format depends on the channel. Keep an approved master transcript and separate caption file where possible. New language or platform versions are much easier to prepare from those files than from a final video alone.
Which videos need subtitles?
Whenever speech carries important information, plan a text version. Interviews, explainers, recruitment clips, webinars, product introductions and social videos all fall into this group.
Review videos without speech as well. A short text cue may be necessary when music and images alone are expected to explain an action. Relevant sounds can be described for Deaf and hard-of-hearing viewers. The legal accessibility duties that apply depend on the organisation, service and intended use. Obtain advice on the specific case when a binding assessment is required.
Subtitles also help in ordinary situations: on a train without headphones, in a noisy shop or when someone can read a language more confidently than they understand it in speech. Accurate text makes specialist terms, numbers and names less ambiguous.
Plan for captions before filming. A contributor who speaks very quickly, uses long nested sentences or is frequently interrupted off camera creates more editing and timing work later. A concise spoken structure benefits both the audience listening and the audience reading.
Turn a transcript into readable subtitles
Automatic speech recognition can remove much of the typing. It cannot make the editorial decisions. Listen through the full video and compare each line with the recording.
Prepare a list of words that are easy to mishear:
- Names of people, companies and products
- Towns, streets and regional terms
- Specialist vocabulary and abbreviations
- Figures, prices and dates
- Words from another language
- Similar-sounding words with different meanings
A word-for-word transcript is not always a useful subtitle track. Speech contains repetitions, filler words and abandoned sentences. Shorten with care while preserving the meaning, tone and intention. If a pause carries meaning, do not let the text reveal the next idea too early.
Build lines around natural phrases
Break lines at natural language boundaries. Avoid separating an article from its noun, a first name from a surname or a verb from the phrase it completes. Two short, logically divided lines are usually easier to follow than one dense block.
Use normal punctuation. Captions are not telegrams. Full stops help readers separate ideas. Ellipses should indicate a genuine interruption or continuation, not appear automatically at the end of every subtitle.
When several people speak, make the change clear if the picture does not do so. Descriptions such as [door closes] or [quiet music] belong in the caption file when the sound contributes to information, action or mood. They do not need to list every noise in the room.
Keep terminology consistent across the whole film. If a product or department has an agreed spelling, record it in the project glossary. Alternating between two versions can confuse viewers and makes later translation harder.
Check timing and presentation
Text should appear as the statement begins and remain long enough to read. A caption that arrives early gives away the next answer. One that arrives late forces the viewer to reconcile sound and text from different moments.
Watch the full captioned version at normal speed. Reading only in an editing interface with the picture paused gives a false impression of comfort. If a line cannot be read easily, shorten the wording or divide the caption again. Speeding up the entire video is not a solution.
For burned-in captions, pay close attention to typeface, size and contrast. The background changes from shot to shot. A restrained background panel, shadow or outline can keep text legible. White lettering on its own will disappear over a pale wall or window.
Leave room for platform controls, logos and important action in the image. Captions placed without a preview can cover a product, a speaker's hands or a face. Test the actual post or advertising preview on a small phone and a larger screen.
Legibility is more than type size
Large letters do not help much when every word jumps into view in another colour. Animated word-by-word captions may suit a particular short format, but they add visual movement and make quality control harder. A steady presentation is often easier to follow when the subject contains technical detail.
Use emphasis selectively. If alternate words change colour, the hierarchy disappears. The sentence should remain understandable without decoration.
Separate file or text in the export?
Files such as SRT and WebVTT store text and time codes. Many video and web platforms can use them as an optional track. This makes corrections, search and multiple language versions easier. Check which format the destination platform currently accepts.
Burned-in captions remain visible on social media exports even when the platform offers no separate track. Viewers cannot turn them off, however, and every language requires its own video export.
Often the practical answer is to keep both: a clean master video, an approved subtitle file and channel-specific exports with text in the picture where needed. This keeps the original material available for later changes.
Use filenames that identify the project, language, format, version and date. A folder filled with files called final_new_2 makes the next correction slower and increases the chance of publishing the wrong copy.
Translate without losing the meaning
Subtitle translation needs context. A short sentence can have several meanings when the translator cannot see the picture or understand what came before. Supply the video, transcript, approved spellings and notes about the intended audience together.
Literal translation is rarely the best fit. Languages differ in word order and sentence length. The target text must fit the available time while retaining the important information. Technical terms, brand names and forms of address need an agreed treatment.
Ask someone who knows the target language and understands the subject to review the finished version. Automated translation may provide a draft, but publishing it without a check risks incorrect terminology, tone and timing.
A separate subtitle file normally gives more flexibility across several languages. Dedicated social exports may still be needed where the platform does not offer a selectable track in the chosen format.
Use a dependable approval process
Captions deserve the same care as names, prices and figures shown in the picture. A short production sequence catches most problems:
- Generate speech recognition output or a raw transcript.
- Correct it against the recording.
- Shorten spoken language and divide it into readable lines.
- Ask the responsible expert to check names, figures and terminology.
- Test timing at normal playback speed.
- Review presentation on target devices and platform previews.
- Archive the approved language file and final deliverables.
The speaker should not be the only proofreader. Someone who knows what was meant often reads the intended wording rather than the text on screen. A second person is more likely to spot a misspelt name or missing word.
Teams that publish video regularly can maintain a small glossary of brands, people, locations and specialist terms. It speeds up speech recognition checks and human review without turning each project into a fresh terminology exercise.
Frequently asked questions
Do subtitles automatically double video reach?
No. There is no sound basis for a general promise of doubled reach. Subtitles make videos usable in more situations, while distribution still depends on subject, opening, audience, platform and promotion.
Are automatic captions enough?
They are often a useful draft. Before publication, check language, names, figures, timing and line breaks. Technical vocabulary raises the error risk.
Which is better, SRT or burned-in captions?
SRT is flexible and can be disabled when supported. Burned-in text remains visible everywhere. Keep a clean master and separate caption file whenever the production allows it.
Should captions describe sounds?
Describe a sound when it carries information, changes the action or contributes to the intended mood. The legal requirements for accessibility should be checked for the specific service and use.
Can spoken sentences be shortened?
Yes, provided the meaning, tone and important information remain. Remove filler words and repetition first. Do not rewrite a claim to make it sound more certain or more promotional.
How can we prevent mistakes in names?
Prepare an approved spelling list before transcription, compare the text with it and ask the relevant person or subject expert to check the final version.
Want captions and platform versions planned from the start of the shoot? OzyCore Studio's video production service in Wiesbaden brings recording, editing, text and export into one production plan.