AI video translation · dubbing · subtitles

Put your video in
9 languages.

Paste a link or drop a file. Minutes later you have a dubbed, subtitled master in every language you publish — in your own cloned voice. Like hiring a localization studio, a hundred times faster.

★ The part nobody else does: fix one line, re-render only that line try it below ↓

Start free — 5 min on us ▶ Try the interactive demo

Credits never expire · Your footage auto-deleted in 7 days · Commercial license · No credit card

Live UI demo · switch languages, click any line, edit the text, roll a new take — fix one line and only that line re-renders.

app.voxtakes.com / projects / zelos-x9-launch · Spanish

Line review · 5 lines

QC via re-transcription

Mixdown unlocks when all lines pass · this job used 12.4 / 60 min

Free tier exports carry a corner watermark · removed on any paid plan

Real output

Hear it for yourself.

A 33-second clip of a real NASA scientist interview, dubbed end-to-end by our pipeline — cloned voice, timed subtitles, background music kept. Pick a language, press play. Public-domain source (U.S. government work), so you're hearing our output on footage anyone can verify.

Grading = per-line machine re-transcription QC on this exact clip. No stock footage, no cherry-picking.

Why we exist

Black-box dubbing makes you pay for their mistakes.

Today's tools treat a finished video as a one-shot generation. But real work always has that one line that's off — the question was never whether you can edit it, but what it costs to do so.

One word fix = full re-render

Regenerate the entire job. On per-minute tools that means paying for 20 minutes of credits because of one mispronounced brand name — plus another render wait.

Voxtakes: only the edited line is re-recorded, and QC re-dos are on us

Old captions fighting your new ones

Re-uploading a subtitled cut means two caption tracks stacked on the picture. Removers charge for it or the fix is pixel-cloning by hand in an NLE.

Voxtakes: burned-in subs erased in-pipeline, then fresh styled captions burn in

Credits that expire at midnight

Monthly credit pools reset whether you used them or not. Creators publish in bursts — flat monthly plans punish exactly that rhythm.

Voxtakes: credits never expire, top-ups valid for 12 months

New in the studio

The layers other tools make you outsource.

Schematic previews below — every one of these is live in the product today.

Hard-subtitle removal

Erase the old captions before dubbing

OCR finds the burned-in text, an inpainting model repaints the real background underneath — clothing, water, studio walls all reconstructed. Your new captions land on clean video instead of ghost text. One pass serves every language.

Before · burned-in subs video frame with burned-in Chinese subtitles

+$0.24 / finished minute · skipped automatically when the source is already clean

Caption presets

One render, platform-ready looks

Pick a style at upload and every dubbed language ships with captions burned in that look — or export the SRT/ASS files and keep the styling in your NLE.

CLEAN

La Tierra observada desde el espacio.

YouTube / documentary

BOLD POP

¡25 SATÉLITES MIRANDO LA TIERRA!

Shorts / Reels retention

WORD HIGHLIGHT

El agua cambia de color

TikTok karaoke pacing

BILINGUAL

Gracias a ellos entendemos el clima.

And that helps us understand the climate.

All four styles included — no separate captioner subscription.

Brand glossary

Your names stay yours in all 9 languages

Lock a term once — the translation model is constrained, and a machine check on every rendered line flags any term that slipped through.

Unlimited terms on Creator and up · per-project lists keep client brands apart.

Multi-speaker + delivery

Two voices stay two people

Voiceprint clustering clones every participant separately and keeps identities bound through the whole video. Choose the deliverable per job:

Host · ASo what changed since the last prototype?Host voice, cloned separately

Guest · BCost. And battery life doubled.Guest voice, kept throughout

Subtitles only · voice dub only · or both. Subs-only runs at half credit.

How it works

Three steps. Nine languages.

STEP 01

Paste a link, or drop a file

YouTube links are pulled server-side at full source quality — nothing to download or re-upload. Local files go in over resumable chunked upload; close the tab, it still finishes.

10-min video ≈ 15-min turnaround

STEP 02

Review line by line. Fix what's off.

Translation and cloned voiceover arrive line by line with QC scores. Play a line, edit the text, roll a new take — each change re-records that line only. Everything else is untouched.

Edit one line, pay for one line

STEP 03

Ship the whole package

Burned-in styled subtitles + SRT/VTT/ASS files + localized title & description drafts. Run all target languages in parallel from one approved script.

1 video × N languages

Pricing

Simple math. Credits that stay yours.

Metered in finished-video minutes (rounded up). Edits are charged for changed lines only. Every plan's credits roll over forever.

MonthlyAnnualSave 20%

Starter

Weekly creators

$29 / mo

60 dubbed minutes · ≈ $0.48/min

9 target languages, parallel outputs
Line-by-line review & per-line billing
URL pull / resumable chunked upload
Burned styled subs + SRT/VTT export
Credits never expire

Get started

Creator

Full-time YouTubers & podcasters

$79 / mo

200 dubbed minutes · ≈ $0.40/min

Everything in Starter
Voice cloning — keep your timbre & delivery
Multi-speaker separation
Brand glossary & term lock
Hard-subtitle removal · caption presets
1080p masters · no watermark

Start free — 5 min on us

Studio

Agencies & localization teams

$199 / mo

600 dubbed minutes · ≈ $0.33/min

Everything in Creator
3 seats · shared workspace & approvals
Client review links
API & bulk queues (beta)
Usage reports · configurable retention

Talk to us

Subtitles-only render half credit (≈$0.20/min) · burned-in removal +$0.24/min · overage $0.89 / finished minute · failed-QC re-dos never billed

Compare

Run the numbers against the big names.

Real scenario: a 20-minute master that needs 2 line fixes after review.

	Voxtakes	Rask	ElevenLabs	Subtitle-only tools
First 20 min	$9–16($0.40–0.48/min in-plan)	≈$60–150($2.4–3.0/min)	≈$44(Dubbing v2, $2.20/min)	≈$4(subtitles, no voiceover)
Cost of fixing 2 lines	2 lines only · ≈$0.04per-line billing, no re-render	Full re-render in web UI ≈$48segment redubbing is API-only	◐ Studio supports per-clip regeneration	Back to the NLE by hand
Line-by-line review + QC scores	✓ Re-transcription QC + approval state	✕ No QC loop	◐ Editor exists, no QC	✕
Burned-in subtitle removal	✓ In-pipeline, one pass for all languages+$0.24/min	✕ Separate manual step	✕	◐ Some offer plain delogo
Caption style presets	✓ 4 presets burned per language	◐ Basic styling	✕	✓ Strong, but no dubbing
Brand-term glossary	✓ Lock + per-line enforcement check	◐ Business tier	—	—
Unused credits	Never expire	Reset monthly	Rollover capped at 3×	—

Figures from vendors' published 2026 pricing pages; products evolve — tell us if anything moved and we'll update.

Roadmap

What's landing next.

Beta feedback drives the queue — approved testers vote with their applications.

Lip-sync

Mouth movements re-timed to the dubbed language, so the speaker reads native on camera — the finishing layer over the dubbing pipeline you can use today. In internal testing now.

Soon

Voice library

Studio-grade ready-made voices for when you don't want to clone your own — browse, audition on your script, lock per project.

FAQ

Questions people actually ask.

Create an account — no invite needed. Verify your email, answer a short application (what you'll dub, which languages, must-have features), and a human reviews it, usually within two business days. Approved accounts get 5 free minutes, no card. The open beta is how we pick our first hundred power users.

Yes — the “Hear it for yourself” section above plays a 33-second NASA scientist interview dubbed end-to-end by the production pipeline, on footage anyone can verify. Want one on your footage? Include a link in your application and we may run a line while you wait.

Our masters are assembled from per-line audio tracks on a timeline. Editing a line's text or voice re-generates that line's audio only and drops it back in place — everything else is reused, so billing is that line's duration. No 20-minute re-render for a two-word fix.

Many videos already carry hardcoded captions (from a previous edit, a platform export, a foreign burn-in). OCR locates the text band, an inpainting model rebuilds the pixels under it frame by frame, and the pipeline renders the dub over the cleaned picture — +$0.24 per finished minute, skipped when your source is already clean. Redubs reuse the cleaned master; you're never billed twice.

By finished-video minutes, rounded up. One video dubbed into two languages = two units. Subtitles-only output costs half credit. Edits bill changed lines only; re-dos caused by failed QC are free. Every approved account includes 5 free minutes.

Source files and masters are auto-deleted after 7 days by default, encrypted at rest, and never used to train models. Private object storage, access only via your account. GDPR-compliant processing; DPA on request.

We launched with 9 languages tuned and evaluated line-by-line — English, Spanish, German, French, Portuguese, Japanese, Korean, Mandarin, plus Arabic in beta — instead of advertising "135 languages" with three usable ones. Expect a light foreign accent on the hardest phonemes; we'd rather tell you than promise perfection.

15+ seconds of clean speech is enough. Timbre, pitch range and delivery carry across languages. For long-form coherence we render in blocks rather than isolated lines, so the voice doesn't drift between sentences.

Yes — all paid plans include a commercial license and you own the output. You're responsible for rights in the content you submit; we run moderation and a DMCA process.

Both on the roadmap above — lip-sync is in internal testing, the voice library follows. Beta users who flag them on their application get early access.

Put your video in9 languages.