kyma-dub
Dubs a video into another language with a time aligned voice, in one command.
The problem #
A creator with a good video in one language reaches a fraction of the audience it could. A studio dub is expensive, and subtitles lose the viewers who will not read them.
The approach #
One command chains the steps a studio does by hand: extract the audio, transcribe it, translate the text, synthesise a new voice, and align it to the timing of the original so the speech lands where the speaker's did.
Where it is now #
Public and working today as a command line tool and an MCP server.
What happens if this works #
Language stops being the reason a video only reaches one country.