IBC 2018 Reflections: Microsoft Azure’s Scott Bounds Talks Video Indexer, Power of AI
IBC 2018 marked the general availability of Video Indexer, Microsoft Azure’s (1.C27) AI-based advanced metadata extraction service. Video Indexer – which has been in a public preview since May – provides audio, speech, and visual-based learning models to enable emotion detection, topic inferencing, and facial recognition. In addition, Microsoft demoed how Azure Media Services can help deliver UHD 4K HLG streams from the cloud in partnership with MediaExcel, as well as a “live production in the cloud” demo in partnership with Dejero, Avid, Haivision, Hiscale, Make.TV, and Signiant.
SVG sat down with Scott Bounds, U.S. media and cable industry lead, to get an overview of all the demos on-hand at the booth and to discuss how he expects Video Indexer to aid sports-content creators, how AI is changing the game for sports, the move from on-prem to virtualized infrastructure, and Azure’s variety of cloud-storage options open up new possibilities for broadcasters.
What are some of the big highlights at Microsoft’s IBC stand this year?
The biggest announcement here is really with general availability of Video Indexer across the world. We also have several Video Indexer integrations announced here with Ooyala, Avid, and MAM vendors like Dalet. We’re activating Video Indexer directly with Microsoft and through other [vendors’] platforms so we can get the metadata and those insights directly to the users that need it and integrate it with the tools they use.
We’re also showing a proof of concept [demonstrating] live production through the cloud. We’ve got a live video stream from a Dejero system being sent to Azure, which is handling everything in the cloud – from master control to transcoding to recording and everything else. It’s a good proof of concept for a news organization or a sports organization that all the primary workflows you do today on-prem can be done in the cloud.
The one thing you will really see from Microsoft Azure at this show is that we’re not just focused on the backend distribution side, we’re really concentrating on content creation and distribution, as well.
We also feel we have an advantage when it comes to working within the partner ecosystem, because we are a non-competitor platform. We’re offering a highly scalable platform around the world, but we’re also not competing against you like some other major cloud vendors. We are not a media company, we are a technology-platform company.
How do you foresee Video Indexer and other machine-learning tools having an impact on sports-production workflows?
You’re seeing AI impact sports and media in a lot of different ways. With Video indexer, you’re looking at augmenting metadata in content so that you can unlock details that you don’t have today. We also see AI replacing human tasks like logging because [Video Indexer] can automate the workflow. So we believe AI is going to streamline and improve efficiencies in people’s pipelines.
We see AI [use cases] ranging from simple things like transcribing a video interview and inputting it into your editing platform to more advanced things like how Endemol Shine uses [Video Indexer] for its Big Brother [production]. They use speech-to-text and facial recognition index video files. It cuts down on hours and hours of work and allows them to focus on the creative side of things.
And then you’re also seeing AI used as a recommendation engine and as an insights tool for how to actually monetize the content. For the content-creation side, intelligent storage and management is going to be big so you actually know what is in your archive by using AI. If you have a deep library, AI is going to let you find and see things you would have missed otherwise.
On the monetization side, just think about the potential for personalization. We have a great partner – ZoneTV – that built their platform on top of Azure and Ooyala. They are using AI to pull information on the content and create a personalization service that can automatically give you content from all kinds of different [online outlets] based on your preferences and what you’ve watched in the past.
We are definitely still in the early days, but businesses are already finding efficiencies and new ways of doing things using AI. It’s about baby steps right now, but people really see where this can go in the future and how AI is going to impact your whole content supply chain.
More and more broadcasters are looking to virtualize pieces of their production workflow in the cloud. What role do you see Microsoft Azure playing in this process?
The demo we have here with live-video streams going into the cloud is on the bleeding-edge side. But I think what’s really coming on strong in the short-term is being able to edit totally in the cloud without any client desktops on-prem.
I talk with sports organizations all the time that ask, “how can we put our video in the cloud, and then access our whole archive, edit our video with a workstation running up in the cloud?” Well the answer is we can do that today. And it’s not just on a web-browser-based editor, we can actually run a full Avid Media Composer or Adobe Premiere Pro system up in the cloud. We have people now who are doing this today and aren’t buying any on-prem hardware. So that’s a very big shift.
How is Microsoft Azure’s cloud-storage offering evolving to meet the needs of broadcasters?
The foundation of all of this is really what we are doing with Azure storage. We take a flexible approach to storage with Blob [object] storage. We have different temperatures of storage ranging from Azure Archive Storage for low-cost storage to premium storage, which offers high-throughput. And once you’re inside Azure you can automatically take advantage of all the intelligence offered by Video Indexer.
Also, once you’re inside of Azure, you can [take advantage of] Microsoft’s network, which is the second biggest fiber network on the planet. So if you’re in an Azure datacenter – and we now have 54 regions incorporating multiple datacenters – you can take your content around the world and it all stays on Microsoft fiber. So if I have some content that’s sitting in the east coast of U.S. and I want to ship it to western Europe, we’re not going out over the internet; we’re using Microsoft fiber for that. So really that hyper scalability is really key for us.