One upload, five AI stages, a dub that holds up to scrutiny
Every video that passes through FaceMagic moves through the same disciplined pipeline — built to hold up on close-ups and fast dialogue, not just a demo clip.
The same clip, before and after
Play one, then the other. The voice you hear in Hindi is the speaker's own, and the dub runs to the same length as the source — which is what lets the lip-sync hold.
AI-dubbed demonstration — the person shown did not record or approve it.
Transcribe & diarize
Isolate every speaker and transcribe the original audio, timestamped line by line.
Translate & localize
Neural translation (built on AI4Bharat IndicTrans2) with gender-aware re-translation for grammatically correct Indic output.
Re-time for lip-sync
An isochrony rewrite re-times the translated speech to match the original's duration, frame for frame.
Clone voice & generate speech
Voice-cloned text-to-speech reproduces the original speaker in the target language, not a generic replacement voice.
Sync & finalize
Audio-video alignment and automated lip-sync match the on-screen mouth to the new language before export.
Nine Indian languages, one dubbing pipeline
English source video in, voice-preserving dub out — across India's major languages, including:
Studio dashboard, or a single API call
Use the FaceMagic studio for hands-on review, or call the API directly from your pipeline to dub renders automatically the moment they land.
- Webhooks for render-complete events
- Captions & subtitle files included
- On-device processing available for sensitive content
curl https://api.orturalabs.com/v1/dub \
-H "Authorization: Bearer $ORTURA_KEY" \
-F video=@episode-04.mp4 \
-F target_language="hi" \
-F clone_voice=true
# → returns a project id + webhookCompared to the alternatives
| Capability | FaceMagic | Traditional dubbing studio | Generic AI/TTS tools |
|---|---|---|---|
| Speaker's original voice preserved | |||
| Grammatically correct Indic output | |||
| Automated lip-sync | Rare, manual | ||
| Turnaround per video | Hours, not weeks | Days to weeks | Fast, but low quality |
| Cost per language | Fraction of studio cost | High (studio + voice artists) | Low, but visibly synthetic |