A peek inside the Machine Intelligence Garage

It’s exciting to feel the energy and passion from start-ups condensed into brief, focussed pitches delivered directly by the experts. The blood, sweat, and tears that go into the product are palpable and listening to three excellent pitches from Digital Catapult’s Machine Intelligence Garage cohort was no exception.

I was invited to take part in a recent event with the cohort, along with the DPP’s CTO, Rowan, and other DPP members Codemill and the BBC. The 11 innovators in the Machine Intelligence Garage’s Creative Industry focussed cohort all have incredible value to bring to media companies, broadcasters and studios – it was only a shame we had time to hear three.

First up was envsion.io – tackling the problem of generating relevant metadata to power video search. This has been an elusive challenge for media organisations who have had to balance all kinds of trade-offs when deciding how to make video searchable. Do they curate a proprietary dictionary of terms, people, and objects they care about, then record where they appear in the video? Do they try to make sense of a firehose of descriptive data from off the shelf image recognition tools? Envision propose a solution, and we’re looking forward to seeing how they progress, to see if this can eventually become a problem of the past.

We all know that feeling of finishing your favourite series only to discover it’s going to be years before the next one.But Charisma.ai think they have an answer. They pitched a toolkit that lets storytellers (not coders) generate interactive experiences so viewers can continue to spend time with favourite characters (and coincidentally maintain your relationship with the content brand) before the next series comes round. The approach to help the storytellers do this themselves was not only empowering and accelerating, but makes the solution particularly scalable and deployable.

The final pitch of the day was from DeepReel.com. The emergence of global streamers has undoubtedly advanced audience excitement for globally sourced content. But for various reasons, preferences for dubbing versus subtitling have formed differently in different regions. DeepReel could be about to disrupt that, and regions that typically prefer subtitles - like the UK and Nordics - might be about to get more dubbed content. The solution presented uses AI to modify the original actors’ lip movements so that they match the dubbed dialogue, without many hours of intricate post production work. The solution delivers a convincing experience, and gives dubbing artists a greater choice of language beyond the scope of the original lips, enabling them to stay closer to the original story.

According to Cognizant’s own research, the media industry is behind in adoption of AI based technologies compared to other industries. Yet it has close to the highest level of ambition for AI based technology – expecting to be a leader in only three years. To realise this ambition in this timeframe needs significant acceleration. That’s why both Cognizant and the DPP are keen to support innovation in this space.

Identifying scalable opportunities to deploy and realise value from AI technologies requires deep collaboration between business teams, technology teams, and partners. So it’s rewarding to see these start-up cohorts not only blazing a trail in technology terms, but also engaging deeply with the media industry.

We’re looking forward to participating in further DPP research this year on the use of AI and automation in media supply chains. But if you’d like to learn more about the state of AI in the media industry, you can download our report, AI in Media & Entertainment: starting the journey to value, and the DPP’s report, AI for Real.