How to Make Narakeet Videos in 12 Languages at Once

November 17, 2026

You can produce one Narakeet video per language from the same script by setting the language and voice on the first line and rendering again — twelve exports from twelve one-line changes. The catch is the visuals: on-slide text stays in the source language unless you split it out. Narakeet supports over 90 languages and roughly 800 voices across its tiers.

Plan for that split before you write a word, and the multi-language run takes an afternoon instead of a week.

How do you switch the language of a Narakeet video?

Change two lines at the top of the script:

language: es-ES
voice: Lucia

Render. The narration is now Spanish with the same timing structure. Repeat with de-DE, fr-FR, ja-JP and so on. Narakeet picks reasonable voices by default if you only set the language, but named voices keep consistency across a series.

Native voices exist for major languages; some smaller languages get a single voice and noticeably flatter delivery. Test one 30-second sample before committing to a full series.

What breaks first in a multi-language build?

The slides. Any text you typed onto a slide is an image of English, and no amount of voice switching fixes it. Two fixes:

  • Keep slides text-free. Use shapes, colour blocks and icons; put all words in the narration. Renders identically in every language.
  • Use image slides per language. If you must show text, build a set of slide images with text baked in, one per language, and reference them with image lines.

Option one is less work and performs better on phones anyway.

How do you keep timing and captions aligned?

Because the script drives timing, a longer Spanish translation shifts every slide. Translation expansion is real: Spanish and German run 15-25% longer than English for the same content. Either shorten the source lines (so translations stay inside a comfortable length), or accept a video that is ninety seconds longer in German.

Captions render per language automatically. Download the SRT for each so you can edit names and product terms that the engine mispronounces.

Does translating the script yourself beat machine translation?

Yes, for anything customer-facing. Machine-translated scripts produce grammatically correct narration with odd rhythm, and the TTS engine reads the odd rhythm out loud. A human pass over a 600-word script costs less than re-rendering four times.

For internal training, machine translation plus your own proofread of the first minute is usually good enough.

When is this workflow the wrong call?

If you need lip-synced human presenters, this is not the tool — look at avatar platforms. If your content is heavily visual with on-screen data, per-language slide sets double your asset work and the ROI drops. And if you only need one language, the multi-language planning here is wasted effort; just write the script and render it.

Quick Summary

  • Change language: and voice: on line one, render again — one video per language.
  • Slide text does not translate; keep slides text-free or build per-language images.
  • Translations run 15-25% longer, so timing shifts between languages.
  • Human review of the script beats MT for customer-facing content.
  • 90+ languages, but smaller languages may have only one voice.

Voice and language coverage is on the Narakeet tool page.