Runs entirely on your machine

Dub any video into 9 languages — without uploading a single frame.

Voxlira reads the video, erases burned-in subtitles, re-records every line in your own voice, and lands each sentence on the original timing. No per-minute billing, no queue, no cloud.

Download for macOS
Requires macOS 12+ · Apple Silicon (M1 or newer) · 8 GB RAM
Download for Windows
Requires Windows 10/11 (64-bit) · DirectX 12 GPU · 8 GB RAM
Free plan: first 60 seconds of any video Pro from $8.3 / month Full system requirements
Nothing uploadedFiles stay on disk
No minute limitsUnlimited renders
9 languagesOne click each
Keeps the timingLines land on cue
Why local

Cloud dubbing bills you by the minute.
Voxlira bills you once.

Everything — transcription, translation, voice cloning, subtitle removal, the final render — happens on the computer you already own.

Voxlira
Typical cloud dubbing
Pricing model
Flat plan, unlimited videos
$10–$60 per minute rendered
Your footage
Never leaves your disk
Uploaded to someone else's server
Turnaround
Runs at GPU speed, no queue
Queued behind other jobs
Burned-in subtitles
Erased and re-typeset automatically
Left on screen, or cropped away
Sentence timing
Each line fitted back to its own cue
Drifts out of sync over long videos
Offline use
Works with Wi-Fi off
Requires the internet
What's inside

Built for the messy videos you actually have

Hardcoded subtitles, multiple speakers, tight shot clocks, footage you're not allowed to upload — Voxlira was designed around those constraints.

Subtitle eraser

Finds the burned-in text, removes it only on the frames where it appears, then typesets the new language in the same spot — no black bars, no crop, no ghosting.

pixel-accurate mask · feathered edges

Your own voice, in every language

The voice reference is collected from the whole video, so the dub keeps your timbre and cadence instead of sounding like a stock narrator. Separate speakers stay separate.

per-speaker voice profiles

Nine languages, one click

English, Mandarin, Japanese, Korean, Spanish, French, German, Italian and Russian. Sentences are translated, then spoken — never the other way round.

untranslated lines are blocked, not faked

Timing that survives the edit

Each new sentence is fitted back into its original slot within about a tenth of a second per changed character — and never pushed so hard it turns to noise.

clear speech first, speed second

Preview before you commit

Scrub the timeline, nudge a single line's start or end, re-listen to just that sentence, fix a translation typo — all before a single frame is re-encoded.

per-sentence audition & re-render

Private by architecture

No upload endpoint exists. Client footage, unreleased ads and NDA material can be processed the same way as public videos — because nothing is transmitted.

runs after first-run model download
How it works

Three steps, one export

No prompt engineering, no timeline assembly, no manual lip-sync pass.

Step 01

Drop the video in

Voxlira detects the speech, splits it into sentences with exact timecodes, and shows you the transcript in a timeline you can edit directly.

Step 02

Pick a language and check the lines

Every line is translated and marked if it still needs a human look. Fix wording, adjust a cue, or re-listen to one sentence until it sounds right.

Step 03

Render and keep the master

The original voice is replaced, subtitles are re-typeset in the new language, and the finished file is written next to your source video.

Before you download

Will it run on my machine?

Voxlira runs models locally, so it needs a real GPU-class machine — not a thin laptop.

Minimum

SystemmacOS 12+ / Win 10+
CPU4 cores
Memory8 GB
Free disk15 GB
GraphicsMetal / DX12

Recommended

SystemmacOS 14+ / Win 11
CPU8 cores or more
Memory16 GB+
Free disk30 GB+
GraphicsApple M-series / RTX

First launch

Model download~4.7 GB
Typical peak memory~7 GB
30 s test clipunder 2 min
After downloadFully offline
Devices per license2

Not sure? The app checks your hardware on first launch and tells you exactly what's missing.

Pricing

One flat price. Render as much as you want.

Every plan includes all 9 languages, the subtitle eraser and unlimited exports. No credits, no metering — a license key is emailed the moment you pay, and each key activates 2 devices.

Monthly
Flexible creator
$19.9/ month
Cancel anytime

For one-off projects and trying the whole pipeline end to end.

  • Unlimited video length
  • No watermark on exports
  • Up to 4K 60 fps
  • 2 active devices
Get monthly key
Quarterly
Active creator
$39.9/ quarter
≈ $13.3 / month · save 33%

For channels and accounts that publish every week.

  • Everything in Monthly
  • Multi-speaker dubbing
  • Background music preserved
  • 2 active devices
Get quarterly key
6 months
Studio sprint
$59.9/ 6 months
≈ $9.9 / month · save 50%

For course libraries, podcasts and agency production runs.

  • Everything in Quarterly
  • Batch-friendly workflow
  • Commercial use allowed
  • 2 active devices
Get 6-month key
Best value · save 58%
Annual
Unlimited pro
$99.9/ year
Just $8.3 / month

The whole studio, paid once a year. This is what most people buy.

  • 365 days of unlimited dubbing
  • Full commercial licence
  • Priority engine updates
  • 2 active devices, both platforms
Claim annual pass

Prices in USD, charged by Creem. Your key arrives by email and is redeemed inside the app.

FAQ

Questions people ask before buying

Does my video get uploaded anywhere?
No. Transcription, translation, voice synthesis and the final render all run on your own machine. The app only talks to the internet for account sign-in, licence checks and the one-time model download (~4.7 GB on first launch). After that you can work with Wi-Fi off.
How does the free trial work?
Download the app, sign in, and you can process the first 60 seconds of any video with the full pipeline — eraser, voice clone, subtitles and all. The trial output is physically cut at 60 seconds; buy a key and the same project exports in full.
How do I receive my licence key?
Payment is handled by Creem. As soon as it clears, a key like XXXXX-XXXXX-XXXXX-XXXXX-XXXXX is emailed to you. Open Voxlira, click the Pro badge, sign in and paste the key. Your account is upgraded instantly.
Can I use one key on two computers?
Yes — each account covers 2 devices at the same time, for example a MacBook and a Windows desktop. You can unbind an old machine from your profile at any time and free the slot.
What hardware do I need?
macOS 12+ on Apple Silicon, or Windows 10/11 64-bit with a DX12-capable GPU. 8 GB of RAM and 15 GB of free disk is the minimum; 16 GB and 30 GB is comfortable. The app runs a hardware check on first launch so you know before you start a long render.
What if a translation is wrong?
Every line is editable. You can fix the wording, re-translate a single sentence, or nudge its start/end time and audition just that line. Lines that fail to translate are flagged and blocked instead of being silently replaced with the original audio.
Which languages are supported?
English, Mandarin Chinese, Japanese, Korean, Spanish, French, German, Italian and Russian — in any direction, including dubbing a Japanese video into Spanish. Japanese input is read as kana so kanji compounds aren't pronounced with Chinese readings.

Try it on your own footage

The free download processes the first minute of any video — no account required to install, no card to start.

Download for macOS
macOS 12+ · Apple Silicon · 8 GB RAM · 15 GB free disk
Download for Windows
Windows 10/11 64-bit · DirectX 12 · 8 GB RAM · 15 GB free disk
Download for macOS · free

macOS 12+ · Apple Silicon (M1+) · 8 GB RAM · 15 GB free disk