A client sends a three-minute training film and asks for it in eight Indian languages by next week. It sounds like one job done eight times. In practice it is four different jobs.
1. Adapt the script, don’t just translate it
Hindi takes about the same time to say as English. Tamil and Malayalam often take longer. If the translated line runs past the speaker’s mouth, the dub looks wrong. So each script is adapted to fit the timing of the picture, line by line, before anyone records.
2. Cast for the audience, not just the language
A Gujarati voice for a bank sounds different from a Gujarati voice for a cooking channel. We send two or three samples per language so the client can pick, and we keep the same artists for the next film so the brand sounds the same every time.
3. Change what is on the screen
Titles, captions, lower thirds, charts and phone screens in the video are all in English. For a film to feel local, those change too. This is the step most often forgotten, and the one viewers notice first.
4. Mix and check every version
Each language gets its own mix, and each one is watched end to end by a native speaker before it leaves us. That last watch catches the mispronounced product name and the subtitle that covers someone’s face.
Where AI voices fit
For training and internal films, AI voices now make sense: they are quick, cheap to update when a policy changes, and consistent. With a speaker’s written consent, we can even clone their voice so a founder speaks in languages they don’t. For ads and anything that needs real feeling, we still cast people.
If you plan a film for many languages, tell us at the script stage. A few small choices early, like leaving room in the timing and keeping text off the picture, save days later.