WinCAPS Script Extractor: A Quantum Leap In Processing Efficiency

In an industry replete with technical standards, it is bizarre that one important area of information interchange is still apparently a free-for-all, writes Andrew Lambourne, Business Development Director at Screen Systems.

Scripts – be they pre-production or post-production – come in a bewildering variety of different layouts. There appear to be getting on for 200 different ways of setting out speaker identity, dialogue, elapsed time and production narrative, depending on which company produced the script and for whom.
Anyone wishing to make use of such content in a machine-readable form can waste a lot of time trying to isolate just the material they are interested in.
A subtitler serving deaf and hard-of-hearing viewers needs to produce a programme transcript in the same language as the dialogue. Using a script could save a lot of time, but much of the benefit is eroded if they have to spend 20 minutes fiddling in the word processor trying to delete all the other unwanted material. Even if they create macros to do the job, the likelihood is that the next script is in a completely different layout.
Similarly, a translator will be presented with explanatory material or markup identifying onscreen captions in a whole variety of layouts and presentation styles. Broadcast scripts appear to be designed only to be read by humans, and not by machines. In a world where XML is almost de facto for exchanging other types of textual and numerical content, script production has a lot of catching up to do.
Screen Subtitling Systems specialises in making workflow processes more efficient. For subtitling or dubbing customers who rely on using scripts to increase their productivity, there was an imperative to speed up the process of extracting useful payload as quickly as possible from any script layout.
Such payload might include speaker identification labels, dialogue, timecodes or feet-and-frames, scene boundaries, onscreen captions and explanations to assist translation. Faced with the challenge of identifying and isolating these useful information types from a range of script layouts, and discarding unwanted material, the Script Extractor was created.
Script Extractor is an assistive tool, designed to enable a user quickly to identify examples of the material of interest, and to automate the task of retrieving all similar instances from the script. Essentially the script is loaded and broken down into its constituent elements. The user highlights one or two items of each of the content types they are interested in.
Using a sophisticated pattern matching algorithm, the Script Extractor then picks out similar content to be preserved, and ignores the other items. The user can quickly check the results, adjust if necessary, and save the payload into a format useful for their work. A process that used to take 20 minutes or more can now be completed within a minute or two.
If a series of scripts are received from the same production company, all in the same layout, then a 'template' can easily be saved so that the 'training' provided in one session can be retrieved and used to speed up the extraction of the content from the next script of that type. This makes the process even faster.
The output may be imported into the Screen WinCAPS Qu4ntum subtitling system – where speaker labels can be used for automatic subtitle colouring, and where the dialogue and timecodes are used to create subtitles. Alternatively, the extracted content can be imported into a translation editing system where localisation is performed. Given that both subtitling and translation are fiercely competitive businesses where productivity is vital, the system pays for itself very quickly. And until the industry recognises the futility of having such a myriad of different script layouts and moves to adopting an XML standard for script and metadata content, the Script Extractor is a vital tool in the post-processing armoury.

Read the article in the online edition of RFV here.
(IT/JP)