Half-localized support content is still support debt
A walkthrough dubbed into Portuguese with English captions and English callouts does not read as localized — it reads as unfinished. Viewers trust the frame more than the voice, so any English left on screen becomes the thing they notice.
Support content is where that matters most, because the person watching is already stuck.
One pass, every layer
Capture localizes the asset as a set rather than a stack of exports: the dubbed narration, captions, titles, and every bit of text in the video move together into each target language. There is no round trip where audio is translated in one tool and burned-in text is rebuilt by hand in another.
That also keeps timing honest. Translated sentences are longer or shorter than the original, and the caption and callout timings are regenerated to match the dubbed track instead of drifting against it.
The article and the video say the same thing
Every help centre eventually contains a video that says one thing and an article that says another, usually because they were localized months apart. Because both are generated from the same translated script, they ship together in each language and change together on the next edit.
Built on the same localization stack
The dubbing and cloning here are Braiv Dubbing — full-video AI dubbing and voice cloning — with caption work shared with Braiv Captions. The narration voice comes from Braiv Speech. Capture adds the support-specific parts: cleaned instructional wording, on-screen text handling, the replayable capture, and the written article.
That also means the choice of source voice is yours at dub time: clone the original speaker, or clone the polished Speech voiceover that cleanup produced.
Where this sits in Braiv Capture
This is the localization attribute inside Braiv Capture. It runs after voiceover cleanup so translation starts from a clean script, and pairs with automated localized screen recording when the interface in the frame should be localized as well.
