Efficient multilingual dubbing of movies based on existing sub-titles
Original Publication Date: 2004-Dec-06
Included in the Prior Art Database: 2004-Dec-06
Described is an efficient process by which to transform a pair of movies, one with sub-titles and one without, into a dubbed version using OCR (Optical Character Recognition) and text-to-speech technologies.
Efficient multilingual dubbing of movies based on existing sub -titles
From the beginning of talking movies, there has been a constant need for translating the spoken contents from one language to another. The ideal procedure is, of course, to perform dubbing, which is, however, quite an expensive procedure; it requires involvement of professional actors who are trained in following the lip movements of the actors performing in the movie. An alternative more affordable solution was to imprint sub-titles containing the translation. As a result there, is now a huge corpus of movies in which the translation is presented as sub-titles.
Described here is an efficient process by which to transform a pair of movies, one with sub-titles and one without, into a dubbed version. The main idea is to perform OCR (Optical Character Recognition) on the sub-titles and, using text-to-speech technologies, produce the dubbing. The huge advantage is that this process can be made almost entirely automatic, thus substantially reducing the cost of dubbing movies. The proposed process is much faster than the standard dubbing process and can allow very fast international distribution of movies. This type of automation can enable a new, automatic dubbing service for film studios and can open the door for dubbed movies in countries that still lack this service as a standard.
The invention is based on the following process:
1) In this first stage, the two movies are synchronized, and the fra...