Euromedia sep oct sep oct - Page 26

Story “However, the benefits of AI go far beyond just automation.” - Stuart Almond, Sony Professional Solutions Europe criteria. With user data, that will get even better, you are able to personalise cover artwork based on what is more likely to appeal to a specific user. The same is applicable to content itself.” “It absolutely can,” agrees Child, “and it could potentially bring about more timely and more relevant content to viewers, either by matching up more precisely what content a viewer is interested in, or making it easy to find. In the medium term I also hope to see meaningful set of data that they can use to create richer, more compelling storytelling.” “Absolutely, yes,” says Mukherjee. “AI will be content- aware, so it can go a long way in content curation and helping schedulers quickly identify the right content. This may require enhancing existing workflows to AI-assisted ones. Interra Systems is developing technologies for automatic classification and tagging of content using machine learning.” “Once broadcasters can acquire all of the sources they need in real-time, they will face a fundamental challenge that only AI will be able to solve: it would be a treasure trove of content but an overwhelming amount of video to manage,” suggest Jacobi. “Using AI, broadcasters will automatically filter and develop refined levels of metadata to accelerate the curation and programming of all of these new live feeds. Now they’re even more of a feedback loop from viewers into choices of what content is being made, which will further improve those choices. “Yes,” says Sherwood. “This is a major focus area for commercial AI companies who want to help broadcasters monetise the mountains of content that exist in broadcaster’s archives, both old and new. Examples include up-sampling old content for higher resolution devices, auto-tagging content for driving higher relevancy to viewers and advertisers, and even auto-generating schedules for programming.” “Many broadcasters have very large archives, and manually searching millions of hours of programmes would be simply impossible,” admits Karp. “With the move to IP technology and the cloud, these archives become increasingly digitised, migrating the content in the right formats for the application of ML and AI. “The recent example is BBC Four’s experimental AI and archive programming. AI helps programme makers and schedulers unlock valuable resources and put together a manageable selection in a fraction of the time. According to BBC, ‘this enabled researchers and schedulers to uncover programmes they may never have been able to find, and will broadcast a selection that haven’t been seen in years’.” ARCHIVE. “AI will help reduce the number of mundane tasks in the broadcast space so that staff members can dedicate more time to creative, impactful ones,” suggests Martin. “More than that, this emerging technology will deliver enriched information that will make it easier for them to do their jobs. Through tools like facial and voice recognition, content creators and producers have access to a more able to provide dynamically created, sport- like TiVo’s entertainment technology graph— specific, team-specific and athlete-specific live that can automate the production of content channels, interweaving authentic, hyper-local metadata—are key to providing an enhanced live coverage, creating a whole new level of set of information that can be used to open up tailored viewing experiences.” the archive and fuel new business models.” “In theory of course, after you collect the “AI engines with object and voice data you must harmonise it,” says Guillemot. recognition already exist to automate the “However, it is very difficult to do, so, in process of tailoring clips to the appropriate theory it could work but it would take an distribution methods such as streaming to enormous amount of work. Docherty is cell phones, tablets, television and social another in the ‘Yes, definitely’ camp. “AI can media channels,” advises Shen. “The audience be used to understand the content better, and and the presentations differ, depending on provide information about what types of users the content and where it’s being viewed. will be interested in what specific content. Additionally, AI can make this process simpler It can also be used to determine when this through creating metadata to archived content is most likely to be watched, which can material that does not contain the granularity be used to inform scheduling.” of metadata as well as creating data for live LEGACY. “It certainly could do this,” says video in real-time. Right now, producers Morgan. “One trend we are finding at the hand-craft the multiple versions required for moment is broadcasters looking to monetise a TV show. In the ‘smart studio’ AI engines archives by making legacy content available will be able to automate the assembly of the over-the-top. There are trends where the material and deliver it in the most effective young generation are suddenly discovering way to the target audience.” old programming and AI could certainly HUMANS. In terms of scheduling, it has help surface other items in a series or similar been said that humans have nuance, taste content.” and judgement – the qualities of which no “There is a clear role for AI in mitigating algorithm or machine can replace - yet. To what the huge complexity of We should programming,” states question if Almond. “It can take up to 18 months from the inception we want AI of a creative idea for a to determine programme to be broadcast, 100% what during which time the we want to climate in which the show is watch.” – being broadcast in may have changed drastically. One Petr Peterka, way to ensure the content Verimatrix remains relevant is ][\BUTQQPBRH[\ܝ[XZH\Hو\^]H[[\\]H]\]\K[[™X\[ۜ˸'B'\]X[KRH[[Y\\XZB[HوH\[[[وYXH]H^B[XYH]H]Z\\[ \]H[H[[\Y[\YYX[\ܘ[[Y\[Y[[\Y\[ܙX\H][[X[H\Xܜ][HۜY\Y[XZ[Bܘ[[YH[[]\[]HY]YY[H]HY[YK'B]\XK]Y\\YY\[\[\H]Y\[ۈ\H^H[™\[H\XK8'HY\\ 8&]\H\[\[YO]\H[Y\X[[YO\H^HY[RH\[\ۂH[Yۏ&x'BT Xܙ[]\ۙH[[B܈Y\\\XX[H[\[[HXY[\وZ\\]K8'[H\ Z\H[Z]YY]Y]H\ܙY]۝[ [][H[[ܛ[\\X[[]Y]Y]H\^x&\[\H[ XY\