Portable film subtitler

The subtitler is conceived as two distinct, but related, products: one for domestic use, the other for cinemas. Both types of the machine would have applications for the deaf and hard of hearing and for watching foreign films supplied (whether to the cinema or on DVD) without subtitles; the domestic version could also be helpful in learning foreign languages by supplying the text of what is being said (a slight variant on those for the hard of hearing, which tend to be abbreviated) and could in theory be used with videos and TV programmes as well. Assuming it was made easy to use and was not too unattractive to look at, this should make for an internationally in-demand product.

Domestic version

The domestic version easily seems the more straightforward. The subtitler would be a tiny, dedicated computer – in the same way that video game players or iPods or mobile phones are computers. A computer that does little more than run QuickTime and send the movie to a short, wide screen to show the subtitles is, I imagine, going to be a ridiculously simple piece of kit – more straightforward than the average calculator, or at any rate a mobile phone: indeed, it might be worth working with an amateur computer builder to come up with a prototype. (As well as asking relatively au fait friends about this, I have actually approached the university engineering department and a couple of the computer engineering companies in the Cambridge area. Typically, they didn’t tend to understand what I was trying to get at, although one did say that I should try to find a software engineer, as all the other components were readily available.)

Curiously enough, the screen turns out to be a bit of a deal-breaker. The obvious choice as far as I was concerned was a small LCD monitor: however, very few of the friendly ‘experts’ I asked considered this first, coming up with seemingly more complicated ways of doing it (eg ‘piggy-backing’ the signal onto the TV screen, projecting the titles onto the screen). When I finally ran into a computer engineer who could see the point of this idea, I immediately put my foot in it by saying that I thought that the subtitler should be no more than £30. He pointed out that the cheapest I was likely to get a screen for was £20, and that I should be looking at a price of around £200. So that was that. LCDs are, of course, horribly expensive. But both PLEDs and E-Ink are apparently much cheaper to produce. It might even be worth investigating whether to create a prototype using E-Ink’s prototype kit (www.eink.com/kits/), whose 3.6" x 4.8" (91 x 122mm) display panel would serve fairly well as the subtitle screen. The kit costs $3,000 and already has a Linux-based CPU and standard sockets (USB, etc): the various other things required of a subtitler prototype (receptivity to remote control, small navigational applications) would presumably not be too expensive to develop – quite a lot less, in other words, than the £25,000 quoted to me earlier this year. It might be worth finding out how much the unit cost for a particular quantity of E-Ink-based screens would be.

The titles themselves would be text files, using the "QuickTime Text Track" format, which could be downloaded from a specialist website and transferred to the subtitling machine, either on a memory stick of some kind or (if feasible) by wireless. ‘Text movies’, are they are apparently called, can be created without very much difficulty in any text editing application and then opened and run in QuickTime. They begin with the rubric {QTtext}, followed by some 23 lines of information, each likewise in curly brackets, about how the text should look and run. The subtitle text appears between timings given in the form [00:00:00.000], which breaks down to – hours:minutes:seconds.fractions of seconds. Because of the simplicity of the language used it is possible to imagine creating a small database, perhaps in XML, that stores the subtitles in a more easy-to-amend form, which can then be exported as a QuickTime file: by separating out the text and the timings, it might make it easier both to alter the timings globally (assuming there is a difference in frame rates or something similar) and to do a preliminary translation of the text using an automatic translation program. (A freeware program called Easy Translator did a pretty good job, as far as I could tell, translating the English subtitles from a QuickTime text file into Italian, though not without meddling with spaces and lines, which took all of a minute to re-apply in a text editor, before the subtitles were returned to the main body of the QuickTime file.) Using a piece of freeware called TitleLAB, you can convert any of the large number of subtitles created for the DivX system – available on sites like subtitles.images.o2.cz, though I cannot vouch for their legality – to QuickTime text files, without the intervention of QuickTime Pro. If there’s a sufficient demand, it’s possible that the DVD distributors would contribute useful information – perhaps even a subtitle template with the times already in. (On the subject of legality: perhaps the most obvious parallel would be with song lyric sites, which have been a very popular resource on the web for many years. However, as this story shows, there have been moves to confirm the illegality of these.)

With QuickTime in place, all the elements for the domestic version already exist: one could set up a small monitor – say, one of those horribly overpriced 6- or 7-inch screens designed to be used in cars – in front of the TV, attach it to a standard PC, alter the QuickTime file’s settings to take up the whole screen (many of the required features – a much wider range of playing speeds, the option of removing everything except the subtitle box – are found in a shareware application called QTViewer, created by Daniel Hazelbaker: www.blueboxmoon.com/software), and run the program – taking care, of course, to synchronise its start with that of the DVD. This is pretty clunky, of course, but it clearly works. In some areas it would make more sense to run the file from a standard computer onto a specialised monitor (in a cinema, for instance, where the PC could be run from the projecting booth), but in others it would better to have the thing as a complete entity – something that could be attached to a PC much as a digital camera or a PDA can be now. The data – ie the particular set of subtitles needed – can presumably be supplied on some form of memory stick, or even by wireless, and started by remote from the projection room. As for the domestic monitor, the default shape produced by the TitleLAB conversion – 640 x 40 pixels, equivalent on a 72 ppi screen to 8.9 x 0.5 inches (227 x14 mm) – seems adequate, though it would probably be worth doubling the height (a ratio of 8:1 for the width to height seems more or less okay for both domestic and cinema use). There is no real need for colour, although hard-of-hearing subtitles often use them – and it’s possible that anyone offering ‘value-added’ subtitles, perhaps at a small cost (the 99 cent/79 pence per song on iTunes is a useful comparison), would want to use colour in some way, along with other formatting elements (italics, bolds, smaller and larger fonts).

There remain some hurdles, of course, the most obvious being that the subtitles would be running independently of the DVD. This is also part of its advantage, and shouldn't matter unless something goes wrong with the DVD. (Though clearly if there were some obvious way of getting timing information from the DVD, it should be taken advantage of...) There are plenty of other potential problems, though many of them, I suspect, would present the properly-informed engineer with little trouble.

Producing a remote that can control both the subtitling machine and the DVD player simultaneously should be a relatively easy task, since it is already possible to walk into an electronics shop and buy so-called ‘universal’ remotes. Rather than coming up with two separate remotes, it might be worth considering whether the existing DVD remote could be used to work the subtitler – for example, by altering the subtitler’s firmware. Different set-ups for different makes of DVD player (and therefore remote) could be downloadable from the website. This could also in theory take account of the minor differences between DVD players – the non-standard speeds of ‘fast forward’ and ‘rewind’, for instance – so that the subtitler would do the same things as the DVD player. Alternatively, in terms of convenience of use, it might make more sense to have everything working through a dedicated subtitler remote if that proves possible: it could have a screen like an iPod or mobile phone, and many subtitles could be kept on it and catalogued in a straightforward, readily-accessible way. (The iPod’s various ways of calling up songs, with tools that seem no more than up, down, left and right buttons, might be a useful reference.) It would send the relevant subtitle to the screen by wireless, presumably.

Fixing the subtitler to the TV (given the various shapes and sizes TVs come in) might need some thought, but a certain amount of trial and error will presumably come up with the best solution. Early thoughts: assuming the screen were relatively flexible, it could be supplied with a very basic ‘cradle’ to fit into, which could be stuck directly onto the TV above or below the screen. One example would be a long piece of flexible plastic connected to two narrow rectangles of plastic, which have an adhesive on one side (to stick to the TV) and grooves along their inner edges, allowing the screen, which would terminate at each end with shapes that fit into the grooves. (The grooves could be replaced by something else that would allow the screen to ‘click’ into place: in any case, the idea would be that the adhesive rectangles would not need to be as tall as the screen itself.) The ‘cradle’ could be put at the top of the TV, with the screen hanging down, or at the bottom, with the screen sitting on top of it. The principle behind these various options is to be able to remove the screen safely, without having to twist or otherwise damage it when removing the subtitler from the TV. It would also be worth removing the subtitler when it is not in use, but keeping the cradle in place. Would it also be worth making the subtitler in different widths? (One assumption we have to make is that there are now adhesives that stick things firmly together, but can then be removed without very much trouble. These may not exist. What alternatives would there be? Would the screen be sufficiently light to allow some form of suction to work? Have a look at the pads on the Vacucom site.)

Alternatively – and especially if it had been decided to go with the ‘clever remote’ option above – it might be worth pursuing an option brought up by some of the people I spoke to: ‘piggy-backing’ a signal showing the subtitle text straight onto the TV screen, using the same principle as the volume symbols or teletext produced by the TV remote. (This of course complicates the software and simplifies the hardware, as you would no longer need the extra subtitler screen, but would probably not be able to use QuickTime and would need some potentially complicated way of generating the words. It seems to me that the complication here is a bigger deal than the simplification, but I could very easily be wrong…)

As for dealing with the different speeds of DVDs from various regions (I understand that PAL runs a bit slower than NTSC), there are ways of altering speeds in QuickTime, so this could be solved at the playing stage: it could also be done, and perhaps more efficiently, at the QuickTime-generating stage. (Again, the website would need to be pretty impressive: in addition to carrying many thousands of subtitles, plus potential firmware (covering all the main DVD players on the international market) for the remote, it would need to have specifications for many DVDs, so that the created QuickTime track would be usable in specific countries alongside specific DVD versions of the film. Does this suggest that the track would be generated on the fly? Would it better to assume that anyone who is sufficiently interested in getting hold of the subtitler)

As for the actual generation of the subtitles in the first place, between the DVD distributors who would have a potentially large new market with relatively little outlay, and the enthusiasts who have already put tens of thousands of subtitles onto the web (via the DivX system mentioned above), this should have a certain amount of ready-made product. Creating subtitles from scratch can be done by a judicious combination of DVD player (typically on the computer) and text editor, although it is easy to imagine a simple dedicated application that sits on top of, or otherwise incorporates, the DVD player software and allows the user to ‘mark’ the moments when a character speaks, replay the player between specific points to re-hear the words, and write the translation with the time automatically noted. This application (which should be free, presumably) could be downloaded from the website where the already-compiled subtitles are found.

Cinema version

The cinema version is a much taller (should that be wider?) order: the screen required for this appears to be, in monitor terms, very large indeed. (NB A rough estimate of the sort of size a cinema subtitler would need to be – say, 8 by 1 feet (96 x 12 inches) – gives us a total pixel number, assuming 72ppi, of 5971968 pixels (6 megapixels).) But it seems that the general principle of using a screen (albeit very large in this case) to show a very simple QuickTime movie is much the same as with the domestic version above. It would also have to be positioned in such a way that it doesn't overlap with the cinema screen - perhaps a little forward, though not so much that viewers couldn't focus on both screens at once. Perhaps some of the newer screen technologies would be a better bet than the conventional LCD screen: for instance, Gyricon (who already have a product called SyncroSign, which seems to be on the same sort of track), or E-Ink’s Segmented Displays, or the pleds/oleds being developed at Cambridge Display Technologies. (Gyricon and E-Ink, using segmentation, produce fixed-width characters, which may prove more difficult to read.)

(Further news: Gyricon, as this web link suggests, seems to have given up the unequal struggle against the more business-savvy E-Ink.)