Your film, now in six more languages at once

A good film takes care: the right scenes, the right words, the right pace. Starting over for every market throws that care away and invites drift, because the German version slowly stops saying what the English one said.

What people wanted was the same film, in another voice and another language, with the structure they had already approved. So dubbing starts from the existing storyboard, not from the website again, and creates one new film per language, from one to six at a time.

Translating a narration sounds simple until you look at what is inside it. Some words belong to someone, some marks belong to the author, and numbers follow local habits. So before the text goes to the translator, three things are set aside and put back afterwards, exactly:

• quoted lines stay word for word, because a quote is someone's words
• pauses the author wrote stay exactly where they were placed
• figures and dates are rewritten in the target language's own way, so 1,500 in English becomes 1 500 in French

Each language then starts like any other film. Asking twice for the same language does not start it twice, and images already made for the original are reused, not generated again.

A French voice over English screens only half convinces. When the website has a translated version of a page, declared the standard way for search engines, the filmed scenes are recorded again on that translated page. The French film then shows the French site, menus and buttons included.

When there is no translated page, we reuse the original footage and the scene report says so plainly. We would rather tell you that scene four still shows the English page than let you find out in front of a client. You decide whether that matters, or swap the scene.

A new language often needs a different delivery, so the voice got finer controls in the same release. Each scene can now carry its own voice, speed and emotion, and we still join everything into one voice track, so the captions stay in sync with every word.

Choosing a voice in a language you do not speak is guesswork, so every voice now comes with a spoken sample, 72 in all. Your assistant can play you a few before you commit a whole film to one.

A pronunciation list fixes the words a voice gets wrong: a brand name, an acronym, an unusual spelling. It changes only what the voice says, never the captions, so the screen keeps the real spelling.

And you can now write a pause, from a tenth of a second to three seconds, right into the narration. It becomes real silence, is left out of the captions and the word count, and is never trimmed away as if it were dead air. A dramatic beat stays a beat.

The release also adds Arabic, Hindi, Korean, Polish, Turkish and Swedish. We did not add a language by flipping a switch. For each one we generated sample narration and listened back twice: once with a speech recogniser, to check the words come out as written, and once with an AI listener, to judge whether it sounds natural.

Arabic needed more than a voice. Its captions read right to left, so the caption engine learned to lay them out that way.

Japanese and Chinese passed the same listening test, and we are holding them anyway. Our captions break lines between words, and those languages do not put spaces between words. Rather than ship captions cut at random points, we will add them when the captions can do them justice. The full list of languages is on the features page.

Dubbing is available today through the tools your AI assistant uses with PageToVid: pick a finished film, name the languages, and each version arrives as its own film to review, edit and download. Every existing film works exactly as before.

If you have not connected an assistant yet, the assistants page shows how in a couple of minutes.