What is video translation?

Video translation is the process of converting the spoken and written language of a video into another language, through subtitles, a translated voiceover or full dubbing. It changes the language a video uses, but not its examples, visuals or cultural references.

Reviewed by Clark Quinn, Chief Learning Strategist Updated October 28, 2026

Example

A pharmaceutical company’s L&D team has a library of English employee onboarding videos. New hires are now joining in Brazil and Japan, and they need to understand the content in their own language from their first day. There is no budget to remake each video.

The team translates each one. The narration is translated and re-voiced in Portuguese and Japanese, the on-screen text in each scene is translated, and translated subtitles are added for anyone watching without sound. The visuals, examples and structure stay the same. A local HR partner in each country reviews the translations before the videos go live in the LMS.

How it works

What gets translated

A video carries language in three places, and a full translation covers all of them:

• Spoken audio: narration, dialogue and interviews.
• On-screen text: titles, labels, text in slides and diagrams, and captions burned into the image.
• Text around the video: subtitles, the title and description, and any quiz questions or buttons in an interactive video.

Ways to translate the audio

• Subtitles: the original audio stays, and translated text appears on screen. This is usually the fastest and cheapest option.
• Voice-over translation: a translated voice plays over the original, which stays faintly audible underneath.
• Dubbing: the original speech is replaced entirely, often with lip sync so the speaker’s mouth matches. AI dubbing does this with synthetic voices, sometimes cloned from the original speaker.
• Regenerating the video: for videos made with AI avatars and text-to-speech, the translated script can be used to generate the video again in the new language, with new narration and matching mouth movements.

The right method depends on the video. A presenter talking to the camera benefits from dubbing with lip sync, because viewers watch the face. A screen recording or slide-based video works well with a translated voiceover, since there is no face to match. Subtitles are a good addition to either, for viewers who watch without sound.

The workflow

1. Transcribe the original audio, or start from the script if you have it.
2. Translate the script, using a list of product names and terms that must stay the same in every language.
3. Have a fluent reviewer check the translation, especially technical, medical and legal terms.
4. Produce the translated audio, subtitles and on-screen text.
5. Watch the final video in each language. Check the timing, text that overflows its box, and whether names are said correctly.

Machine translation and AI voices have made steps 2 and 4 fast. Steps 3 and 5 still need people.

Video translation vs video localization

The two terms are often used as if they mean the same thing, but they do not. Video translation changes the language of a video. Video localization adapts the whole video for a market: it includes translation, and it also changes examples, visuals, units, date formats and cultural references that do not fit the audience.

A translated safety video might still show a US-style electrical outlet and quote costs in dollars. A localized version would replace both. For many training videos, translation is enough. Localization is worth the extra effort when the content depends on local context, such as regulations, customer scenarios or humor.

Common mistakes

• Forgetting text inside images. Screenshots, diagrams and burned-in captions stay in the original language unless someone recreates them.
• No term list. Without an agreed list, the same product feature can end up with three different names across a course.
• Skipping the native-speaker review. Machine translation reads fluently even when it is wrong, so errors are easy to miss without a fluent reviewer.
• Letting versions drift. When the source video changes and only some languages get updated, regions end up training on different content.

Why it matters for L&D teams

• Learners follow training in their own language. Training in a second language takes more effort to follow, especially for technical or safety content.
• One source, many versions. Translating an existing video costs far less than producing a new video for each region.
• The same message everywhere. Every region gets the same content, which matters for compliance and policy training.
• Updates have to reach every language. Each change to the source creates work in every language, so choose a workflow where updates flow to all versions together.