When detection is complete, the translator / adaptor takes over. This, along with the acting, is the most challenging and artful part of the dubbing process. The adaptor uses the detector’s indications to translate the original, selecting words that synchronize as well as possible with the movement of the screen actors’ lips6. Culturally specific references, idiomatic expressions, and plays on words are the most difficult to adapt, making the adaptor’s craft a highly specialized one. The key to a successful adaptation is timing and ease of delivery. If the adaptation is well done, an actor will have no trouble reading the text smoothly and with good sync at the time of recording.
This being said, even the finest adaptation spoken with the best possible sync by an experienced actor will look “dubbed”. Editing the recorded dialogue can tighten up the sync, although with the typical time constraints for most projects, there are limits to what can be reasonably expected. Future technological developments may bring about interesting changes to this part of the dubbing process.
This is how a phrase from the feature animation film Coraline was translated and adapted from English into French and Spanish.
Note the blue lines indicating where and how the adaptor has chosen to sync the translation to the original: both coincide with closings of the on-screen character's mouth.
When an actor is known to be in a scene, but the mouth is not visible, he can be made to say anything. This is a trick used by adaptors to include additional details about a scene, or to fit in a phrase that would otherwise seem too long and contrived to comfortably match the original.
When the mouth moves, something must come out. In original language, when unvoiced hesitations, lip smacks, and other mouth movements are inaudible, they go unnoticed because everything else that comes out of the actor's mouth is in sync and completely believable. In dubbing however, much like in animation, movement equals sound, and things look wrong when a moving mouth on screen does not produce a sound.
The detection or adaptation processes can advance at the rate of about 20 min. / day, which is roughly the length of a film reel.