"This town's like a great big chicken, just waiting to get plucked." That line is one of the more unintentionally funny results of cleaning up 1983's notoriously blue Scarface for cable, and new insight from Disney Research could make awkward redubs like that a relic of the past. By using an automated system that generates alternative dialog while keeping the spoken words in sync with lip movements, Walt's mad science wing thinks it has they key to believeable audio replacement for movies and video games -- perfect for anime and foreign flicks, we'd imagine. For example, Disney says (PDF) that the phrase "clean swatches" is replaceable with "like swats" or "need no pots," thanks to the lines having similar phonemes (the smallest form of speech that differentiates two words, like "bat" and "bad").
By combining those with dynamic visemes (visual phonemes), and some old fashioned scouring of a pronunciation dictionary, an acoustic speech synthesizer is able to generate inoffensive audio tracks. Disney says that using this approach, it's able to suggest "thousands" of alternate word strings compared to traditional methods. Whether or not that means we'd actually want to watch an edited-for-TV movie, on the other hand, is another matter entirely.
[Base image credit: Universal Pictures]