|
 Good subtitling, like a good roof, generally goes unnoticed. However, like a leaky roof, bad subtitling can hardly be ignored. Fortunately, good subtitling is easily achieved by qualified professionals following industry standard practices. The following summary of these practices provides the basis for excellent subtitling results. It is based on a contemporary approach using a digital non-linear editing application such as Avid or Final Cut Pro in the production of open subtitles--subtitles that are permanently wedded to the video track of a program. Note that the same basic approach with respect to time-code spotting, script preparation and translation would apply as well in the case of closed subtitling. However, in the case of closed subtitles there would be no character generation or title placement per se, as the titles would be transmitted in a "closed" portion of the bandwidth in parallel with the video track, to be turned on and off by either the end-user or a point-of-transmission technician.
Step 1. Evaluate the material carefully.
More often than not, subtitling is the most economical way to localize a film or video program. Sometimes, however, it is not the appropriate approach. Abundant use of on-screen text often precludes subtitling. Instructional videos in which the viewer needs to carefully watch the action should be voiced-over and not subtitled. On another note, depending on the intended audience, subtitles may not be appropriate for reasons of literacy or cultural habit. Sometimes, a combination of voice replacement and subtitling is the best, most cost-effective approach. If you are unsure of whether to use subtitles or voice replacement, a free estimate from a qualified professional agency may provide the necessary insight.
In the absence of an accurate shooting script, the dialogue/narration must be transcribed prior to translation and subtitling.
This should be done well in advance so that the linguists can review the material. Prior to translation, the script is broken up into title-length segments that are timed against the dialogue/narration.
The translator then translates title-by-title, adhering to the timing of the original. Subtitle translation is a special skill. One must be brief while still relating the essential information and capturing the spirit and register of the original text.
This is the actual creation of the subtitles in which the subtitle editor generates high-quality tiff files. The formatting of the files is adjusted to account for the final format of the production.
This is the marriage of the tiff files to the video. During this step, the video technician/editor will make any final adjustments to the timing and placement of the titles. The titles are then rendered, or digitally combined with, the original video track.
Whenever possible, it is advisable for the client to review the edited draft in house. Any suggested changes will be discussed with the editor and translator, and then implemented.
A qualified linguist reviews the titles for accuracy, timing and typographical errors.
The final program is output to tape or as a digital file.
It is important to note that the exact workflow used to achieve the above steps may be limited to a single entity in the production stream e.g. to a qualified translation agency, or they may shared among multiple stakeholders e.g. a translation agency and a video post-production house.
DVDs can be subtitled in two ways. Either one can render the titles into the original program as described above before creating the DVD, or one can create DVD-specific titles. The former generally affords a higher quality subtitle, while the latter offers the viewer the possibility of turning the titles on and off. In cases in which the client requires subtitles in multiple languages, DVD subtitles are usually the best way to go.
Again the work-flow may be completed in full by a specialized translation agency or it may include multiple providers.
Introduction
This text is generally concerned with digital subtitling of corporate-industrial programming, including that for video in general, as well as for DVD and networked media, in both standard and high definition formats. More specifically, it is a high-level review of a flexible approach to subtitling that lends itself to solving problems that often present themselves in the context of multilingual subtitling for corporate-industrial use, and particularly in the contemporary world in which tape is less and less prevalent as a means of carrying video programming. While it may be of interest to those involved in the practice of subtitling, it is intended as a broad overview for the consumers of such services and other interested laypeople.
The classic approach to video subtitling, one that is still commonly used, is that in which a proprietary system is used to superimpose titles on a completed program that has been output to tape--digital or analogue. Still employed today though it may be, this approach is outmoded for a number of reasons, and particularly so with respect to the subtitling of commercial-industrial programming. Specifically, it creates a great deal of overhead and burdensome repetition when a program must be titled in multiple languages, when a program is subject to frequent or on-going changes, because one has constantly to re-title an entire tape using the specialized system, for each new iteration of the program.
Current technology permits a more flexible approach to multilingual subtitling in which one subtitle editor or video editor can produce subtitled programs in multiple languages by working with translators unburdened by the requirement of expensive, specialized systems. This approach offers a number of distinct advantages over others: it permits one subtitle editor or video editor to produce subtitled programs in multiple languages working with translators unburdened by the requirement of expensive, specialized systems; it easily accommodates client changes at relatively low cost; it facilitates production of titles for different kinds of media--video, DVD, Flash and even film--even within one project.; and it this offers the possibility of a rationalized work-flow that permits multiple post-production vendors to easily collaborate with each other and with the attendant translators.
The principal technology driving this approach is that of non-linear video editing (NLE). This implies a software or combined software/hardware solution that permits one to edit video (and/or audio) without the limitations implied by the linear structure of tape; one can cut and paste, undo and make all but unlimited changes all but instantly. Secondarily, the approach also requires an actual subtitling software application which facilitates time-code spotting, or the indexing of the in- and out-points of the subtitles, and then the introduction of this timing data and the subtitles themselves into the NLE. While it will include a systematic review of the problems to be solved and the basic solutions to them, this paper will not provide application-specific solutions. Rather it anticipates that such solutions will be sought within the context of a given project's requirements and the resources available to fulfill them. As a matter of basic reference however, the most prevalent non-linear editing systems in use today are Apple Computer's Final Cut Pro and DigiDesign's Avid; there are others, however. There are also innumerable subtitling applications available. Some are stand-alone applications, others function as plug-ins to the NLEs.
Corporate-Industrial Subtitling: Specific Concerns.
Typically, narrative-fiction film and video, and documentary film and video present basic one-off subtitling projects--a movie or television program is subtitled in one or more languages and that's the end of it. The only likely change would be if there were a change in distribution, such that it becomes necessary to subtitle in an additional language or languages.
Along with the necessity of multilingual subtitling, corporate-industrial programming often also presents the more complex issues of re-use, revision, adaptation and re-purposing, as well as that of multiple review by an array of stakeholders. Imagine if, having spent tens of thousands of dollars translating and titling a safety video in a dozen languages, a yearly review by the legal department results in a finding that the video must have an additional five minutes of material and that the phrase "may" has to be changed to "must always be" in ten instances. Moreover, imagine if there are, in fact, five different versions of the program in question, all of which are unique but all of which share a significant amount of material.
The question is how to approach such a project such that:
1. From the beginning, it has been understood that revisions to both text and footage are not only possible but all but definite.
2. Translation and titling done for a recurring portion of the program need only be done once.
3. Extensibility to additional languages is easily attained.
4. The translation and titles can easily be reviewed by multiple stakeholders prior to actual release of the program.
Subtitling Basics
Time-Code Spotting
The first step in any subtitling project is to "spot" the material. This implies somehow noting the run-time in- and out-points of each title using time-code notation. In simple terms, it merely means making a list of those time-code values. Currently, there are dozens of means of doing this and accompanying formats; some are proprietary, others are platform- and/or application-agnostic.
Multilingual Time-Code Spotting
The key to time-code spotting for a multilingual subtitling project is to do it once and then introduce those values into a transcript of the video. This yields what is commonly referred to as a subtitle source script--the source document of a subtitle translation. The script must be such that the titles for which it is a source will fit properly on the screen, and be easily read and as unobtrusive as possible, and typically, it must also preserve at least some of the subtitles' meta-data e.g. title numbers, time-code values.
Script Distribution
The next step is to disseminate the subtitle source script to translators in a "plain vanilla" form--one that requires no special application--that nonetheless ensures that the text strings will conform to the time-code and format constraints. Typically, this means a specially formatted Microsoft Word document--the format of choice for most of the world's translators.
Translation
For present purposes, what is important to note is that the translation is done independently of the preparation and titling steps. This means that the translator can be anywhere in the world and use his/her standard tools. What is also implied is that the translated text strings need not be handled at all once the translation is complete. Rather, upon completion, the translation will be sent back to the video/subtitle editor, who will import it wholesale into the titling system--there is no need for him/her to handle the translated text strings e.g. to cut and past them into text fields in an application or to type them into an application or system.
Subtitling Scripts
The subtitle editor, having received the translated scripts, introduces them into the required subtitling context--whether open, closed or DVD--via a script. This is typically with via specialized software that may be a stand-alone application or may be part of a non-linear video editing application or a DVD authoring application.
Title Superimposition: Video
Contemporary video subtitling benefits from a variety of methods and technologies. The two principal categories are "open" and "closed." Simply put, open subtitles are those that are "burned in" and are thus a permanent part of the image. While the term "open" thus seems somewhat contradictory, note that it refers to the fact that the titles are part of the open component of the video signal. Closed subtitles are those that--like closed captioning*--are carried in a special part of the signal that is closed unless turned on and off by viewers and/or broadcasters with the appropriate technology in place.
Title Superimposition: DVD
Subtitles for DVD can me handled in three basic ways: They can be open (burned into the image), in which case they cannot be turned off and on; they can be closed--to be decoded and turned on and off by the viewer via a television or other device capable of decoding the closed portion of the signal; or they can be DVD-specific titles that are turned on and off within a DVD menu.
A Real-world Example: In-flight Safety Videos
Perhaps the best way to explain how an NLE-based approach saves time and money is to consider its deployment in a real-world situation.
A major airline decides to translate all of its 5-minute-long in-flight safety videos--one each for 8 different aircraft--into 21 languages.The subtitling work-flow begins and ends with a production house charged with the overall project, but they are not accustomed to doing subtitling or translation. They determine that what they want--in order to maximize both control and profitability--is to farm out only that which is outside of their core competencies. Hence, intend to do the final compositing and output. Furthermore, in the interest of economy, they want to avoid the expense of a final on-site review.
Given their similarity, the safety videos present rich opportunity for reusing material. Fully 80% of any one program is identical to that of at least one other program and, in some cases, they differ only by virtue of 10 seconds-worth of material.
The production house turns to a translation/subtitling vendor familiar with the flexible approach envisioned above, one that addresses the four basic concerns: revisability, language extensibility, the ability to leverage identical assets for recurring material, and the possibility for all relevant stakeholders to review the material in a timely manner prior to distribution.
The basic steps of the work-flow are as follows:
0. Production House and T/S Vendor work together to create a unified script that covers all of the programs. This key step means that from the very beginning, repeated footage and text is identified such that subtitles for repeated components of the programs can be leveraged accordingly. When they have completed this essential step, there is one comprehensive video file per language and not eight. So instead of having to title and QC five minutes of material for each language/aircraft combination (8 aircraft x 5 minutes x 21 languages = 840 minutes TRT!), one need only title and check approximately 6 minutes per aircraft (the approximately 4.5 minutes of program material that is shared by all of the aircraft, plus the 1.5 minutes of total material that differs from aircraft to aircraft e.g. emergency exit information).
1. T/S Vendor time-code spots the program. While the production house might have elected to do this themselves, they feel that it is best done by those who are sensitive to the linguistic issues involved. Time-code spotting, while not an intrinsically difficult process, needs to be done with careful attention to grammar and syntax, particularly when many languages and/or many vastly different languages are involved. Furthermore, it is important to note that in the case of corporate-industrial programs and specifically in the case of informational programs, one is faced with two, conflicting constraints: 1) One cannot typically condense the translation as heavily as one would in the case of entertainment programs; and 2) one is typically presented with all but non-stop speech, and often at a fairly high rate.
Note: in most workflows, the T/S vendor need not work from on-line footage. In the case presented, the T/S vendor worked from low-resolution files that were delivered in minutes by FTP.
2. T/S Vendor translates the script. The subtitling source script is properly formatted as a Word .doc file and distributed to the translator for each language. The result is then reviewed and redacted by a second translator. The final files are delivered to the end client.
3. Client reviews the translations. This is an important step to ensure quality in advance of putting the titles on the image.
4. T/S Vendor produces titles/scripts. Using the final, approved scripts, the subtitle house produces the graphic files and/or necessary scripts. In some instances, depending on the applications in use, they may simply forward an NLE application file to the production house.
5. T/S Vendor delivers finalized scripts and/or title graphics to production house. The details of this step depend largely on the application basis of the project. In some instances, the title data is included in scripts, in others, one needs both the script--for timing data--and graphic title files. And in some instances, one may be ably to simply transfer an NLE application file along with title data.
6. Production house produces low-resolution files for final review. A final review by linguists is always desirable if not absolutely necessary in assuring an error-free result. Even if the script translations have been carefully reviewed, it is important to do a final review of the titles against picture. This may be done in situ, but often it is desirable for reasons of cost and convenience to simply forward a low-resolution video file to the linguists providing the review.
7. Production house composites and outputs final titled programs. As has been explained, the final output will vary depending on the applications involved as well as the manner of distribution and display.
8. Revisions. Approximately 9 months after completion, due to equipment changes, it becomes necessary to add 2 seconds of footage and one title in every program--that's a total of 168 programs in 21 different languages. But, because of the way in which a shared title can be cascaded from program to program, the T/S vendor needs only to provide one new title per language, which can then be automatically incorporated into the project in the production house's NLE.
Conclusion
The workflow envisioned above--one taken from actual projects--provides a high degree of flexibility and extensibility. It also presents a case in which only one stakeholder had to use on-line footage--let alone a specialized system, and in which there was no need at all for tape beyond the initial injest into the production house's NLE. In a similar vein, the individual translators--in languages ranging from French to Chinese, and Arabic to Tagalog--could be located around the globe both for translation and quality assurance.
Perhaps most importantly, in the face of a global revision, all of the programs involved could be changed simultaneously. Moreover, should the airline add a new aircraft, the problem of subtitling it can be addressed quickly, easily and at relatively low cost. |
Foreign-Language Subtitling
