For serious immersion learners

Watch the live foreign stream — understand every line, original audio kept.

Babelio overlays your-language subtitles on any native desktop app — lectures, streams, players — while you still hear the original voice.

For investors

No browser extension. Works on the native apps extensions can't reach. We never store or upload your audio.

The wall

You hit the plateau. Native content is the only way through.

Translation lives in silos — Zoom has its captions, YouTube its subtitles, browser extensions cover web pages. None of them reach the native desktop apps where the content you actually want to learn from lives.

01 · PROBLEM

No fan-sub for the live stream

You open a live stream on the native client, catch ~40%, and rage-pause to look words up. By the time you've found one, the moment is gone.

02 · PROBLEM

Pausing every line kills the flow

Rewind, relisten, look up, repeat. The grind feels productive but it breaks immersion — and you give up 20 minutes in.

03 · PROBLEM

Dub it and you stop hearing it

Tools that replace the speaker's voice destroy the one thing you came for: training your ear on real native speech.

The fix

A translation layer under the original — not over it.

Babelio captures the audio of any native desktop app and lays live captions plus a quiet whisper-dub under the preserved original voice — under 700ms behind the source, so it reads as interpretation, not echo.

  • →Original audio stays audible — you keep training your ear.
  • →Works on native installed clients, not just browser tabs.
  • →Full auto-mute dub is there too — one toggle, off by default.
Three steps

One button. No audio-routing setup.

1

Pick the app

Open Babelio from the tray, pick the desktop app you're watching from the live capture list, and choose your language once — it's remembered.

2

Toggle on

Babelio detects speech and streams it through transcription, translation and voice — live captions and a whisper-dub appear under 700ms behind the source.

3

Follow live — then mine it

Follow the whole session in your language with the original still playing. Afterwards, export the lines you want to an Anki card or a bilingual recap.

What's inside

Built for the way you actually learn.

DUAL-TRACK · DEFAULT

Captions + whisper-dub, original kept

A low-volume translated whisper sits beneath the source voice with a confidence-shaded caption — fills the missing 40% without taking your ear-training away.

NATIVE CAPTURE

Reaches apps extensions can't

Per-process audio capture on macOS 14+ and Windows 11 — VLC, MPV, native streaming clients, desktop lecture players. The lane no browser extension can enter.

SENTENCE MINING

One-click Anki export

Every session saves an aligned bilingual transcript. Grab the line you didn't know, send it straight to Anki — sentence mining without the manual copy-paste.

SUB-700MS

Fast enough to feel live

Glass-to-glass under 700ms. If the pipeline lags, it visibly drops to a faster mode rather than drifting into a delayed echo.

PRIVACY

Your audio is never stored

Signed and notarized installer, plain-language permission prompts, audio processed only while a session is on. No screen capture, no keylogging, no upload.

OPTIONAL · AUTO-MUTE DUB

Full dub when you want it

For passive watching or accessibility, flip one toggle to mute the speaker and replace them with an AI dub — and switch back to dual-track instantly.

Why nobody else ships this

A layer between the walled gardens.

Zoom, Teams and Google will keep adding translation inside their own apps — it deepens their seat revenue. But an OS-wide, cross-vendor layer cannibalizes nobody's core, so no single platform is incentivized to build it. The open lane survives precisely because it sits between the walled gardens — and that's exactly where Babelio lives.

Pricing

One plan. Honest by the minute.

$12/mo

Babelio Pro. 7-day full trial, then a free tier you can stay on. Cancel in one click.

Free — 60 subtitle-min/mo, 2 languages Annual −17% — $120/yr
  • ✓~2,000 subtitle-minutes a month — generous and never throttled
  • ✓All languages, all apps, dual-track overlay
  • ✓One-click Anki export and bilingual session recaps
  • ✓300 dub-minutes included; metered overage at $0.06/min
  • ✓We never store or upload your audio
FAQ

Straight answers.

Won't the AI translation make me lazy and stop me hearing the real speech?+
No — that's the whole point of dual-track. The default keeps the original voice playing and lays a quiet whisper-translation and caption underneath. You're still listening to native speech; Babelio just fills the gaps you'd otherwise pause for. Full dub that mutes the speaker exists, but it's an opt-in toggle, off by default.
How is this different from a browser extension like Language Reactor?+
Extensions only see what's inside a browser tab. Babelio captures audio at the OS level, so it works on native installed apps — VLC, MPV, native streaming clients, desktop lecture players — and on live audio where no fan-sub exists. That's the lane an extension structurally can't enter.
Is the translation accurate enough to trust for learning?+
Captions are confidence-shaded — low-confidence words are marked so you know where to double-check, and the original is always one tap away. The translation is constrained to the actual transcribed words, so it doesn't invent content. We gate every model change behind an accuracy and latency eval before it ships.
Another subscription on top of Migaku and Anki — why?+
Those tools handle reading and review. None of them touch live or native-desktop audio — the exact moment you're stuck right now. Babelio is the one piece that lets you watch a live stream at near-full speed and exports straight into the Anki flow you already use. There's a free tier, so you can prove it pays for itself before you pay.
Is my audio private? What about the scary install warning?+
Audio is processed only while a session is running and is never stored or uploaded for your own use. No screen capture, no keylogging. The installer is signed and notarized, so you get the verified-publisher path on macOS and Windows — not a red warning — and a plain-language screen explains exactly why each permission is needed.
Which platforms and languages are supported?+
macOS 14+ and Windows 11 at launch — that's where per-process audio capture is possible without kernel hacks. We start with the 6–8 languages immersion learners need most (Japanese, Korean, Chinese first) and add more as the eval coverage grows. No mobile yet; the defensible primitive lives on desktop.

Stop pausing. Start watching.

Join the waitlist for the early-access build. We're onboarding immersion learners first.