01Before anyone records
Lock the picture and the script. Every change after recording means another session in every language, so it is the cheapest decision you can make early.
Extract the script with timecodes or line IDs, and list the text that appears on screen. On-screen text needs translating too, and it affects how much room the voice has.
- 1. Lock
- Final picture and final script.
- 2. Extract
- Script with timecodes or IDs, plus on-screen text.
- 3. Glossary
- Brand names, terms and numbers, with how each is said.
02Translate for the ear, then check the clock
Translate and adapt the script to be spoken, not read. Then read each line aloud against the picture. Lines that run longer than the shot have to be tightened; speeding up the read is the wrong fix.
Have a native speaker check the adapted script before the session. It is much easier to change a line on paper than in the studio.
- 4. Translate and adapt
- For speech, with the register agreed.
- 5. Time check
- Read aloud against the picture, line by line.
03Record, review in picture, deliver
Record with a native actor directed in the language. Then review the takes in picture, not in isolation: sync, emphasis and terminology are easier to judge against the video.
Deliver the files named to your pipeline, split the way your editor needs them. Agree the naming before the session, not after.
- 6. Record and review
- Directed in the language, checked in picture.
- 7. Mix and deliver
- Loudness to the target, files named to your convention.