From a chain of specialists to one surface
A traditional dubbing chain is scattered across specialists: a transcription house, a translator working against timecodes, voice talent in a booth, an audio engineer rebuilding the mix, a graphics operator re-versioning the lower thirds, and weeks of round trips between all of them. Dubbing folds that chain into a single surface where the source transcript, the timed translation, the generated voices, the stems, and the final picture are all views of the same project. The AI does the heavy lifting at each stage, and nothing advances to the next stage until an editor has had the chance to review and correct it.
From source to dubbed cut
The transcript
Three AI passes prepare the source: speaker-diarized transcription, audio separation into speech and background stems, and on-screen text detection. The result is an editable working document: retime, split, and merge segments, and watch crosstalk in a per-speaker timeline view. Each segment carries a treatment: Lipsync replaces the slot entirely, Dub plays over the original speech as it ducks underneath, CC keeps the original voice, and Discard cuts it from the dubbed timeline.
Translation built for the clock
Each line is generated against its slot duration and the target language's natural speaking rate, so it can actually be spoken in the time available. Expressive delivery tags ride along in the text and steer the voice synthesis. Provenance tracking flags a translation as stale when the source text changes, and flags the audio as stale when a translation or voice changes, so nothing silently drifts out of sync.
Voices and audio
Each segment's line is synthesized in a voice assigned from the project's voices library, then tempo-fitted to its slot. The compile step assembles the full-length audio and produces a master mix plus stems (one per voice, one per CC speaker, and one for the background bed), so the mix can be inspected track by track or taken into an external audio suite. Compiles are versioned.
On-screen graphics
A detection pass inventories every piece of burned-in text with its time window, position, and category, and the inventory is reviewable on the player itself. Each language version translates that inventory, and translated lower thirds render through HTML templates into transparent animated clips, previewed before they are committed to a final cut.
Optional lip-sync
Optionally, an AI model re-animates the speakers' mouths to match the dubbed audio. Generation is driven per stem, with each voice mapped to the picture segment by segment, so the model only touches frames where someone is actually being dubbed. CC and discarded ranges keep their original frames, and long renders are resume-safe end to end.
Review and export
The editor plays the work at every stage: original or dubbed audio against the picture, continuous segment-by-segment playback of the generated takes, and per-stem listening on any compile. Transcripts and translations export to CSV, SRT, WebVTT, and a print-ready PDF dubbing script, with right-to-left scripts handled throughout.
- Arabic
- English
- French
- Persian
Built for multi-language versioning
Dubbing is part of the AI inference support the foundation provides, and it is available to select partners in the MediaGuard program.
Broadcasters re-versioning programming
Carry a finished program to other languages without a chain of outside specialists: the transcript, the timed translation, the voices, the mix, and the graphics are views of the same project, and your own editors review every stage.
Newsrooms carrying reporting across languages
Take reporting to audiences in other languages with broadcast-ready cuts. Per-segment treatments keep statements in their original voice where they must stay authentic, and right-to-left scripts are handled throughout.
Apply through MediaGuard
Dubbing is available to select MediaGuard partners. Apply through the MediaGuard program, tell us about your organization, and we will take it from there.
Apply through MediaGuard