Video captions that are right, readable and everywhere.

Part one of video captions is an accurate transcript. AI speech recognition gives a fast first draft, then a person checks it against the audio, fixing names, product terms, numbers and anything technical. For clinics, schools and regulated topics we use your glossary of terms, and we flag anything we can’t hear clearly rather than guessing. Captions that misquote a dose, a price or a date are worse than none.

Part two is timing and line breaks. Each caption appears with the words it shows and stays long enough to read, lines break at natural pauses rather than mid-phrase, and speaker changes are marked when more than one person talks. Captions for deaf and hard-of-hearing viewers also note important sounds. The W3C’s captions guidance explains these conventions, and its WCAG success criterion 1.2.2 makes captions for pre-recorded video a Level A requirement.

Part three is format. Social videos get burned-in captions in your brand style, sized for phones and placed inside each platform’s safe zone. YouTube and your website get caption files (SRT or VTT) that viewers can switch on and search engines can read; YouTube’s subtitles and captions help explains how uploaded files replace automatic captions. Many videos get both: burned-in for social, files for YouTube and the site.

Part four is translation. Subtitles in other languages are translated for meaning, not word for word, and kept short enough to read at the same speed. A fluent speaker checks every translated file before delivery, and product names, sizes and units are localised where the market needs it. English and Hindi are our standard languages for subtitles and voice-over, and other languages are available on request, depending on what the project needs at the time.

Part five is voice-over, when subtitles aren’t enough. AI voice tools can produce a dubbed narration in another language from the translated script. We use them with care: the voice is disclosed as AI where platforms ask (YouTube’s synthetic content disclosure, for example, covers realistic altered voices), a fluent speaker checks pronunciation and meaning, and we never clone a real person’s voice without their written consent.

Captions also help search and reach. A caption file gives YouTube and your website the full text of the video, which helps the right people find it, and transcripts can be published on the page for readers who prefer text. On social, most viewers start videos muted, so captions decide whether the message lands in the first seconds.

Style matters as well as accuracy. Burned-in video captions should be large enough to read on a phone at arm’s length, high-contrast against any background (a soft box or outline behind the text helps), no more than two lines at a time, and placed where the platform’s buttons and caption bar won’t cover them. We set these rules once in your brand fonts and colours, so video captions look the same across every platform and every editor. Subtitles in other languages follow the same layout, with line lengths adjusted for languages that run longer than English.

Delivery is simple: burned-in versions for each platform, caption and subtitle files named by language, and a transcript. For ongoing work, captions are part of our video editing packages; for existing libraries, we caption in batches, starting with the videos that get the most views.

Captions can be fixed after publishing, too. For an existing YouTube library, we download the automatic captions, correct them against the audio and upload the corrected files, starting with the videos that get the most views. For social videos already posted with weak captions, the corrected burned-in version can be used the next time the clip is reposted or run as an ad.

For regulated topics, such as clinics, medicines or courses with entry rules, we can send the caption file to your team for a final read before it goes live. It adds a day, and it means every term, dose, date and requirement in the video captions has been checked by someone who knows the subject.