48 THE OTT FILES Subtitling
In the early days of cinema, subtitles had to be created line by line and optically composited onto each foreign language version of a film. Ah, those were the good old days, when subtitling was simple, easy, straightforward. Now subtitling and captioning (subtitling referring to translations for foreign language films, captioning to services for the hearing impaired) have become tremendously more complex. Every piece of content must have versions made for every conceivable platform – and requirements for subtitling are increasingly non-negotiable, with many countries adopting stricter laws around accessibility and captioning requirements. The sheer avalanche of online video has pushed subtitling companies to their limits, and new technological solutions and workflows are being used to find new solutions. SDI Media has been a mainstay of subtitling and localisation worldwide. The company has multiple offices across Asia, Europe and the Americas and boasts subtitling services in over 80 languages. Its sectors include everything from film to broadcast to theatre to a variety of digital content. SDI’s CTO Scott Rose recognises the challenges in subtitling in the online world. “Online includes everything from YouTube user-generated content to features on an OTT platform,” Rose says. “For subtitling, the speed in which short-form content needs to be created is the most challenging. In addition, every online platform has its own format, metadata scheme and method of ingest. As the content being localised may have multiple destinations, all platforms need to be properly on boarded to ensure the deliverables meet specifications.” The huge amount of versioning that happens in the online world needs careful consideration. Version control is the first big challenge. “There are editorial versions prior to air that are released for localisation so that the process can get started as an initial pass,” Rose notes. “When the final version is available there is a short window of time
to be made accessible. The use of text- to-speech for audio description is also showing more promise as the technology develops. The initial benefit is that more content could be made available with audio description that might not be otherwise. So while the quality may take a hit at certain moments or with certain languages, the overall impact is more access.” Rose notes that beyond the obvious benefits of machine translation and speech to text, AI can also generate potential benefits from content insight and auto-generation of metadata. Combining AI technologies such as image recognition with speech-to-text and natural language processing can improve the understanding of context. “Language in media is representative of both the visual and the audible, which provide circumstance and causality. When the AI/machine learning technologies can work together to understand juxtaposition and relationships, it improves the quality of the output.” provides localisation services for marketing material around the world for the major film studios. Michael Petch, DMS director of client services, notes that the expansion of video into the online world has expanded the definition of what subtitling is. “Subtitling is traditionally thought of as a process for the deaf or hard of hearing,” he says, “and therefore it has taken some time for larger advertisers to realise the added value of subtitles on their online content. With a wide range of video playing platforms and an ever-expanding range of video sizes and playback functions over social media, it’s quite easy to assume that it’s not worth the hassle to provide subtitles on content. This could be a big mistake for advertisers looking to promote their content to a wider user base, which could include those consuming content on the go, who aren’t able to listen to the audio.” Using subtitles in short-form social media video is quickly becoming standard and it’s forcing companies to make hard KEEPING IT DYNAMIC London-based creative agency DMS
to update the initial pass with changes of the final. The challenge is two fold: how do you make sure you are working on the right version, and how quickly can you discern the differences between the passes and only work on the parts that are changed? “The second challenge has to do with release versions that have editorial differences for an intended audience or a territory. In this case, asset and title controls such as EIDR are very helpful.” New technologies are helping to expand the range of what companies such as SDI can accomplish. The cloud provides an instant network that makes getting material directly into the hands of the subtitlers and captioners much easier, and AI/machine learning tools such as speech-to-text and image recognition are becoming increasingly accessible, simplifying the translator’s role. “The use of speech-to-text and auto- generation of captions has improved in quality and will become more common. The advantage is a much faster turn- around and that will allow more content IT’S VERY EASY TO MAKE BAD SUBTITLES... BUT TOGET IT RIGHT AND TO ENHANCE THE VALUE OF YOUR CONTENT EXPERIENCE IS THE CHALLENGE
feedzine feed.zine feedmagazine.tv
Powered by FlippingBook