AI dubbing on your own PC
Your videos in ten languages. In your own voice.
Demodokos Foundry dubs voiceovers, tutorials, narration and animation on your Windows PC, in the voice of whoever speaks. From $12 a month, with no minutes to count.
7-day free trial. Cancel before it ends and you pay nothing. Then from $15 a month, or $12 a month billed yearly. PayPal needed to start.
Windows 10 or 11. NVIDIA or AMD graphics card.
7-day free trial. Cancel before it ends and you pay nothing. Then from $15 a month, or $12 a month billed yearly. PayPal needed to start.
Windows 10 or 11. NVIDIA or AMD graphics card.
Scenes
Pick a scene. Pick a language.
An AI agent made every dub here with Foundry. Change the language while it plays.
Your ears
Which one is real?
Karen Savage recorded this sentence in Spanish. Foundry only ever heard her read English, then spoke the same sentence in her voice. One of these is her. Your own voice is cloned the same way, from your video.
A
B
“Todos los seres humanos nacen libres e iguales en dignidad y derechos…”
Which one is Karen?
One voice. Every language.
One story crosses all ten languages in one generated voice.
English
Long ago, beside a fire, a singer told a story.
You stay in charge
When a joke dies, fix the line.
“Fun guys” sounds like “fungi”. Translated word for word, the pun is gone, and that is what Foundry's first pass did here. Three rewritten lines bring it back.
AI dub, first pass
Check and rewrite any line
Read every translated line before it is spoken, change the words and generate only that line again. For a language you do not speak, export the lines as a text file and send them to someone who does.
Direct the voice
Give each line an emotion or speaking style: whispered, excited, angry, sad.
Keep names right
Hot Words help Foundry hear names and special terms, and say them the way you want.
How it works
From your file to a finished dub
One guided pass in Foundry. Stop at any step, check the work, and carry on.
- Load your video or audio.
- Transcribe. Foundry writes down every line and who says it.
- Translate each line with the scene around it in mind.
- Clone or choose a voice for each person.
- Match the timing of the original lines.
- Export audio, a finished video or subtitles.
Publishing
One video. Ten audiences.
Every language comes out as a finished video, so you choose where it lives. Foundry has no meter, so ten languages cost the same flat price as one.
- English Original
- Deutsch
- Français
- Español
- Italiano
- Português
- Русский
- 中文
- 日本語
- 한국어
A channel for each language
Export a finished video per language and give each audience a channel of its own.
One video, many languages
Use the sound of each finished video as an audio track on one YouTube video, and viewers pick theirs in the player. YouTube offers this to channels with Advanced features.
Everywhere else
Each dub is an ordinary MP4 for TikTok, Reels, a course or your own site. Subtitles reach 50 languages, as SRT files or as styled captions in the picture.
Privacy
Your footage stays on your PC
Unreleased films, client work and private recordings are the last things you want on someone else's server. Foundry makes the whole dub on your own graphics card, so none of it reaches us.
Local inference. Local storage. Encrypted connection. No creative content uploaded to Demodokos for cloud AI processing.
AI agent
Or let an AI agent do the whole dub
Connect an AI agent such as Claude Code once, then tell it which videos to dub and into which languages. It runs every step in Foundry for you, from the first line to the finished file. Foundry comes with instructions your agent can read.
An AI agent made every dub on this page. This is what it did.
- 14 clips
It transcribed every clip, then corrected the text against subtitles and scripts.
- Ten languages
It read every translation against the original and polished the wording, the jokes and the tone.
- Every voice
It cloned the voices from each clip, or chose a ready voice when we asked.
- 61 dubs
It generated every line and fitted it to the timing of the original.
- 582 lines
It read every line back with Foundry's speech recognition, compared it with the script and redid the ones that came out wrong.
- Finished files
It exported each dub, mixed it and put it on this page.
- Our ears
We listened and gave notes. The agent made the changes.
Included in the Professional plan, from $39.20 a month, billed yearly.
Cost
What the same dub costs elsewhere
Cloud tools count every minute, and count it again for every language. Foundry has no meter. Set your own month and compare.
That is 200 minutes of dubbing a month
Plans read on 7 November 2026 from the ElevenLabs and HeyGen pricing pages, at yearly billing where it is shown. ElevenLabs charges for each language, here at 3,000 credits a minute for automatic dubbing without a watermark. For HeyGen we use 4 credits a minute for audio only, its rate without lip sync, like Foundry. Its Creator and Pro plans stop at 30 minutes per video and Business at 60, and its higher Pro tiers are worked out from its lowest listed price per credit. Extra credits are left out.
Pricing
Dub as much as you like. One flat price.
No minutes, credits or character counter on either plan. The Professional plan adds several voices, longer files, the AI agent and client work.
Pay for the year and save 20%
or go month to month and cancel whenever you like
Your own channel, book or podcast
Creator
$12.00a month
$144.00 billed yearly$15.00 billed monthly
Recommended
Automation and client work
Professional
$39.20a month
$470.40 billed yearly$49.00 billed monthly
7-day free trial: cancel before it ends and you pay nothing. PayPal is needed to start, and the plan you chose bills automatically when the trial ends. Cancel before renewal under Billing.
Enterprise: for teams of 5 or more. Required for 5+ seats, 10+ employees, or $2M+ revenue. Contact our team for custom licensing.
Requirements
What your PC needs
- Windows 10 or 11, 64-bitSSD or NVMe disk recommended
- A graphics card with 6 GBNVIDIA, or AMD with 8 GB (experimental). 12 GB or more gives the best overall Music and Speech performance, including the top music model.
- Internet for sign-in and license checksEvery voice is generated offline.
Not sure what your card has? Press Ctrl Shift Esc, open Performance, choose GPU and read “Dedicated GPU memory”.
Questions about AI dubbing
Does Foundry change the lip movements?
No. Foundry fits the new speech to the timing of the original lines and leaves the picture as it is.
Which languages can it dub?
Any of the ten spoken languages into any other: English, German, French, Spanish, Italian, Chinese, Japanese, Russian, Portuguese and Korean. Subtitles reach 50 languages.
How long can my video be?
The Creator plan dubs one voice, up to 20 minutes per file. The Professional plan dubs several voices, up to 1 hour 30 per file. Neither plan counts minutes across the month.
Do I have to use the original voices?
No. Foundry can clone a voice from the file, and you can give anyone in the video a voice you generated or cloned yourself. Clone only voices you have the rights to.
Can I correct what Foundry did?
Yes. Edit the transcript, the translation, the voices and the timing, generate one line again, and export when it sounds right.
Can an AI agent do the dubbing for me?
Yes, on the Professional plan. Connect an AI agent such as Claude Code, or a script of your own, and it runs the same steps through Foundry's command line and API: transcribe, translate, clone, generate, check and export. Foundry comes with instructions your agent can read. An AI agent made every dub on this page.
What do I get at the end?
A finished video with the background sound mixed in, the voices alone as audio, or subtitles, whichever you need.
Does my footage leave my PC?
No. Foundry does the work on your own graphics card, and no creative content is uploaded to Demodokos for cloud AI processing. An internet connection is needed for sign-in and the license check.
What do I need to run it?
Windows and an NVIDIA or AMD graphics card. NVIDIA with 6 GB or more, 12 GB recommended. AMD with 8 GB through Vulkan is experimental.
Hear your own video in another language
Dub one of your own videos during the 7-day trial, on your own PC, in your own voice.
7-day free trial. Cancel before it ends and you pay nothing. Then from $15 a month, or $12 a month billed yearly. PayPal needed to start.
Openly licensed films, dubbed with AI voices. The people shown do not endorse Demodokos.
Sources and licenses
Sintel: The hut, Dragon lands
(CC) Blender Foundation, durian.blender.org, CC BY 3.0. Changes: excerpts, AI dubbed voices made with Demodokos Foundry, mixed over Blender's soundtrack without voices.
Sprite Fright: Peppermint tea, Fix the line, Cornu aspersum
(CC) Blender Foundation, studio.blender.org, CC BY 4.0. Changes: excerpts, AI dubbed voices made with Demodokos Foundry.
Cosmos Laundromat: Victor and Franck
(CC) Blender Foundation, gooseberry.blender.org, CC BY-SA 4.0. Changes: excerpt, AI dubbed voices made with Demodokos Foundry. This dubbed clip stays CC BY-SA 4.0.
Pepper&Carrot, Episode 6
Animation by Morevna Project and music by Gregoire Lourme, CC BY-SA 4.0, illustrations by David Revoy, CC BY 4.0, source. Changes: excerpt, AI dubbed voices made with Demodokos Foundry. This dubbed clip stays CC BY-SA 4.0.
Die wichtigsten Klimafaktoren
ZDF/Terra X/Story House Productions/Gruppe 5/Luise Wagner, Jonas Sichert, Andreas Hougardy, Rudi Kirschen, source, CC BY 4.0. Changes: excerpt, AI dubbed voices made with Demodokos Foundry.
UV Strahlung
Bundesamt für Strahlenschutz, source, CC BY 3.0. Changes: excerpt, AI dubbed voices made with Demodokos Foundry.
Tango explainer
Simpleshow Japan, source, CC BY-SA 4.0. Changes: excerpt, AI dubbed voices made with Demodokos Foundry. This dubbed clip stays CC BY-SA 4.0.
Sobre la gobernanza en internet
Fundación Karisma, RedPaTodos and Magic Markers, source, CC BY 3.0. Changes: excerpt, AI dubbed voices made with Demodokos Foundry.
Les licences Creative Commons
Ministère de la Culture et de la Communication, source, CC BY-SA 4.0. Changes: excerpt, AI dubbed voices made with Demodokos Foundry. This dubbed clip stays CC BY-SA 4.0.
Which one is real?
Karen Savage, LibriVox, public domain recordings. Changes: background noise of her recording reduced, a soft room tone added under the clone.
Dracula, the tutorial, One breath
Made by Demodokos with Foundry. Bram Stoker's novel is in the public domain.