Your speaker, in every language, in their own voice.
edon.ai clones the voice from the recording you upload, then translates the speech, the slides and the subtitles together. When the content changes, you edit the words and export again. No studio, no re-recording.
- Your speaker’s own voice, cloned once and used in every language
- Speech, slides and subtitles translated in the same pass
- You choose when the viewer sees the speaker and when they see the slide
- Edit the transcript and the voice follows, still theirs
Video goes out of date faster than anyone can re-record it
A rule changes, a product ships, a market opens, and suddenly a video says something that is no longer true. Re-recording it in every language costs weeks and a studio budget, so most teams leave it as it is and hope nobody watches too closely.
The expensive part was never making the first version. It is the Tuesday, eight months later, when one procedure changes and someone has to work out which of the nine language versions still says the old thing.
Subtitles are cheap and half your audience reads them on a phone. Studio dubbing sounds right and arrives in six weeks. A one-click translator handles the audio and leaves your slide deck behind, still in the original language, contradicting the voice.
Updating one module after the content changes
| Comparison | Studio re-dub | edon.ai |
|---|---|---|
| What you change | The whole recording | The transcript |
| Who says the words | A new narrator you book | Your speaker, cloned from the original |
| Slide text | A separate job, separately tracked | Translated in the same pass |
| Adding a language later | Start again | Export again |
One upload in, every language out
Five stages between the file you upload and the video you publish. The one in the middle is yours.
-
Upload
One video, or a whole course in one go. Slide decks are matched to their videos by filename.
- Bulk upload
- Slides matched for you
-
Slice into scenes
The recording is split into scenes, the speech is transcribed, the slides are found and the pauses are mapped.
- Transcript
- Slide detection
- Pause mapping
-
Translate
Every scene is translated together with its slide text and subtitles. Your glossary holds the terms that must not change.
- Speech
- Slides
- Subtitles
- Glossary
-
Review
This stage is yours. Fix the wording, retime the slides and mark the words the translated voice has to stress.
- Edit transcript
- Slide timing
- Word emphasis
-
Generate and merge
The voice is generated scene by scene, in the speaker’s own voice when you clone it, then merged back into one video.
- Voice cloning
- Pacing
- Export
Not right yet? Change it and generate that scene again. Nothing leaves edon.ai until you export it.
Your video. The model that suits it.
No single provider is best at everything, so we do not tie you to one. Which model runs on your project is decided by three things, and you see the result on your own video before you commit to it.
The subject
A plumbing course and a basketball broadcast are not best served by the same model. We use the one that knows the vocabulary of your field.
The language pair
What is strongest out of Lithuanian is not what is strongest out of Spanish. Input and output are chosen together, as a pair.
The audience
An academic module, a shop-floor briefing and a sports channel are three different registers. We match the one your viewers actually use.
Tell us the subject, the languages and who is watching, and we will tell you which models we would run and show you the result on a recording of your own.
What a dubbing studio and a one-click translator both miss
Their voice, not a stranger’s
The voice is cloned from the recording itself, so the Polish version is still your trainer and the French interview is still your presenter. A studio recasts the person. A generic tool replaces them with a voice nobody knows.
Speech, slides and subtitles in one pass
Slide text, subtitles and narration are translated together in the same workflow, so a slide never says something different from the voice, and you choose when the viewer sees the speaker and when they see the deck.
The words you cannot get wrong
Load a glossary and the terms in it survive translation: the product name, the regulated phrase, the word your fitters actually use. Then mark the words the voice has to stress, so a warning lands as a warning.
Edit the words, not the video
Correct the transcript, fix a name, change a line. Subtitles and the cloned voice update with your edit. Nothing goes back to a studio.
Built for anyone who publishes the same video in more than one language
Training is where we started, because that is the business we came from. The workflow is the same wherever a recording, a deck and an audience in another language meet.
Corporate and compliance training
Keep mandatory training current and understandable for every worker, in every language on site.
E-learning and training providers
Sell the course you already own into a new market without re-recording it.
Media and publishers
Reach new-language audiences without booking your presenters back into a studio.
Podcasts
Publish your episodes in more languages, with a transcript you correct before it goes out.
Internal communications
Leadership updates and policy changes that reach every site in its own language, on the same day.
Technical and equipment documentation
Instruction and maintenance video that keeps pace with every language version of the document it belongs to.
Written for the people who have to sign it off
If your training is regulated, the question is never only whether the translation is good. It is who checked it, what they changed, and what you can show an auditor a year later.
- An EU company, built for European rules first
- A Data Processing Agreement available on request
- Our subprocessors listed publicly, and updated before we add one
- Nothing exports until you have reviewed it
Publishing AI-translated video in the EU comes with transparency duties. They fall on you as the publisher, not on your tools. We keep the record of what was produced; the disclosure to your audience stays yours to make.
We make training for a living
edon.ai was built inside a group that produces corporate training in Lithuania. We hit the same wall as our customers: every course had to exist in more than one language, and keeping the versions current was the job nobody wanted.
So we built the workflow we needed, and then we made it a product. We still run our own courses through it, which is usually how we find out first when something is annoying.
Bring a real video
The fastest way to find out whether this fits is to put your own recording and its deck in front of it. Book thirty minutes and we will run your material live rather than a polished example of ours.
Frequently asked questions
What does edon.ai do exactly?
edon.ai translates the spoken audio, the subtitles and the slide content of a video together, in one workflow, with the translated speech generated in the speaker’s own cloned voice. You upload a video and its slide deck, translate them into other languages, and control when viewers see the speaker and when they see the slides.
What does the translated voice sound like?
Like your speaker, if you want it to. The voice is cloned from the recording you upload and used for the translated audio, across every language, with multi-speaker support and word-level emphasis you control. Bring a real recording to a demo and judge it on your own material rather than on a claim on a web page.
Can I lock terminology with a glossary?
Yes. Load a glossary per project or per organisation and the terms in it are held through translation: product names stay themselves, regulated wording stays exact, and the word your people actually use beats the dictionary. Ask a generic translator for a cable tie in Lithuanian and you can get a five-word description of a plastic strap; your fitters say zateškė, and a glossary is how the translation says it too.
Does edon.ai translate both speech and slide text?
Yes. Spoken audio, subtitles and slide text are translated in the same workflow, so you do not need one tool for audio, another for captions and a third for the deck.
Can I control when slides appear instead of the video?
Yes. You decide when the original video is shown and when the slides are displayed. That matters for presentations, training modules and demos, where for long stretches the deck is what the viewer needs to be looking at.
Can I edit transcripts and subtitles?
Yes. Transcripts and subtitles are fully editable, and your change is reflected in the subtitles and in the translated voice.
Can I control pronunciation and word emphasis?
Yes. You can apply emphasis to specific words or phrases to get pronunciation, clarity and meaning right. It matters most for names, technical terms and the sentence someone has to remember.
Does edon.ai support videos with more than one speaker?
Yes. Multi-speaker videos are supported, and each translation stays aligned with the correct speaker.
Can I upload many videos and slide decks at once?
Yes. Bulk upload is supported, and videos are matched to slide decks by filename.
Who checks the translation before it is published?
You do. Nothing leaves edon.ai on its own: you edit the transcript, the subtitles and the timing, and you export when it reads the way you want it to.
Which languages are supported?
Tell us the pair you need and we will confirm it. We would rather answer that question for your actual languages than publish a list that the model layer makes out of date.
How long does a translation take?
Far less than re-recording, and the useful answer depends on the length of the video and how many languages you want at once. Ask on a demo with a real module and we will give you a turnaround you can plan around.
Can I export to my LMS?
Tell us which LMS you use and we will confirm the export format before you commit to anything. Video and subtitle files are exported so they can be loaded into a learning platform.
Where is my data stored and who can see it?
Access is limited to your account users and to the subprocessors we list publicly. We are an EU company and can provide a Data Processing Agreement on request. For the specific hosting region and retention period that apply to your account, ask us and we will confirm them in writing rather than in a footnote.
What are our obligations when we publish AI-translated video in the EU?
They sit with you as the publisher, not with your tools. Our guide to the EU AI Act transparency rules walks through who has to disclose what. We keep the records of what was produced; the disclosure to your audience is yours to make.
How does pricing work?
Three plans. Pay as you go is €7 per finished minute of exported video, per language, on a card, with re-exports free. Outcome is a fixed fee per project: you upload, we deliver the languages checked by a person, and the number is quoted before we start. Enterprise is quoted on volume, with the platform shaped around your organisation.
How do I see it working on my own content?
Ask for a sample and we send a worked example for your use case, or book a demo and we walk through your own material live. Bringing a real video and its deck is the fastest way to find out whether this fits.
Ready to see it on your own video?
Book a demo with your own recording and slide deck, or ask us for a sample first.
