AI dubbing · avatar · localisation

AI dubbing and video localisation into 15+ languages: we carry it through, not just generate it

One video, many markets. With an AI voice clone or an avatar, native lip-sync, reviewed translation and cleared voice rights — so the German, Romanian or Japanese viewer never feels they are watching “a translation”.

15+ languages with native lip-sync
Reviewed translation, not machine output
Voice permissions and licences documented

How is the service different from a dubbing tool?

CapCut, HeyGen and ElevenLabs are excellent tools — we use some of them. The difference is what sits around the tool.

An AI dubbing button leaves three things unsolved. Translation quality: machine translation regularly breaks terminology, brand names and numbers, and sentence length does not match the mouth movement. Voice rights: a CEO's cloned voice cannot be used in a campaign without permission and a contract. Quality control: someone has to check, per language, that subtitles, voice and picture work together.

We add those three around the technology: reviewed translation with native-speaker checking, cleared voice permissions with licence documentation, and per-language QC before delivery. The result is a campaign-ready localised video, not an exported file.

Three methods — depending on the video

Voice clone + lip-sync

For existing on-camera videos. The speaker sounds like themselves in another language, mouth movement adjusted.

  • CEO message, product demo, training video
  • Voice permission from the speaker — we prepare the document
  • Names, numbers and terminology checked by hand

AI avatar narration

When there is no speaker, or they prefer not to appear: an avatar delivers the script in every language, in the same style.

  • One avatar, 15+ languages, consistent brand look
  • Ideal for onboarding, e-learning and FAQ videos
  • Can be part of the social sprint and the video packages

Voice-over + subtitles

For documentaries and B-roll-based videos with no mouth movement. Of the three methods this is the fastest and the cheapest.

  • Studio-quality AI voice or a human narrator
  • Timed subtitle file (SRT) per language as well
  • Finished export for every platform

Languages

Our primary languages are Hungarian and English; for the rest a native reviewer checks the translation.

RegionLanguagesNote
Central EuropeHungarian, Romanian, German, Polish, Czech, SlovakNatural voice sets; Hungarian and Romanian are our own team's languages
Western and Southern EuropeEnglish, Spanish, French, Italian, PortugueseBritish and American English, European and Brazilian Portuguese separately
Asia and Middle EastChinese (Mandarin), Japanese, Korean, ArabicRight-to-left subtitles and cultural check for Arabic

You can ask for other languages too — the list is what we can take on immediately, reviewer included.

Pricing

Localisation pricing depends on video length, the method (voice clone, avatar or voice-over) and the number of languages. One language is typically a fraction of the video's production price, with tiered discounts for several languages. If the video is also made by us (AI video production or the social sprint), localisation can be built into the package — in the video and film production package it is part of the AI post-production.

An exact quote follows within 48 hours, based on the video (or the script) and the language list. Prices are net, in forint or euro.

Process

01

Material and language list

We receive the video or the script, the target languages and the scope of use (organic, ads, internal).

02

Translation and review

Specialist translation, native reviewer, sentence length fitted to mouth movement. Terminology list if you have one.

03

Voice and lip-sync

Voice clone or avatar, with permission in place. A first version per language for checking.

04

QC and delivery

Per-language check of picture, voice and subtitles together, then export in every format with licence documentation.

Frequently asked questions

Frequently asked questions about AI dubbing

How natural does AI dubbing sound?

Today's voice models are natural in every language on the list; the difference is in the text. A reviewed translation fitted to the mouth movement sounds right, a machine one does not. That is why we never leave the translation to a machine.

Do I need permission to clone a speaker's voice?

Yes. Written consent for the use of the voice, with the scope of use, is required. We prepare that document, and it becomes part of the licence documentation.

What happens to names, numbers and terminology?

We check them by hand per language, and follow your terminology list if you provide one. This is the most common error source in machine dubbing.

Do I get subtitles as well?

Yes: a timed SRT file per language, and a burned-in subtitle version if you want it.

How long does it take?

A 1–3 minute video into one language typically takes 2–4 working days including review; several languages run in parallel.

Do you localise videos that were not made by you?

Yes. We need the video in good quality and the voice rights cleared — if someone else made it, we check that localisation is covered by its usage rights.

Send the video and the language list

Within 48 hours you get a quote broken down per language. If there is no video yet, ask for it together with production.