SMW 17: IBM's Scott Grizzle Talks Watson and Video Enrichment
Tim Siglin: Welcome back. Here we are at Streaming Media West 2017 and I have with me today Scott Grizzle, who's a senior solutions engineer with IBM Cloud Video. So Scott, tell me a little bit about what cloud video is from IBM's perspective?
Scott Grizzle: IBM is all under Watson. Watson Media is the overall department head and everyone knows about Watson. So IBM Cloud Video is an acquisition of Ustream and formerly ClearLeap. They merged together and, now with the Watson Integration, Watson Media, Watson Video Enrichment is all one group is IBM Cloud Video and Watson Media.
Tim Siglin: Okay. So what you do mean by video enrichment? That's sort of an interesting term.
Scott Grizzle: It's a way you can bring a video in. Closed captioning versus speech to text. You can do scene recognition. You can do scene detection. You can do a variety of things. It will go through and give you the information with a highlight reel. It will provide you the information that you can work with to create your own sizzle reels, your own highlights, basically do your own editing.
Tim Siglin: So it's more than just metadata. It's more like applied metadata for specific uses. Like for highlight reels.
Scott Grizzle: Correct. It's also creating metadata. So you can put a raw video in with no information, and it'll go through and expand on it. It'll give you keywords. It'll give you themes. It'll give you complete transcripts but it also will create metadata for you that you may not have. Part of the overall enrichment is with the cloud video. We have a logistics manager that allows you to route videos to wherever to pinpoint. You can also do metadata normalization. Right now let's say your iTunes. That meta data is different than when you go to Roku. With Logistics Manager, you can actually go through and make it proper for everything and work with Watson to analyze it and add in more information. It'll package together for you.
Tim Siglin: So it's a rationalization as well as I should say normalization of where it works that. That's interesting. So is IBM's focus with cloud video all sort of geared around Watson and deep learning technologies? Is that where the sort of secret sauce is?
Scott Grizzle: Yes. Again, obviously, with the use of Ustream, you still have the streaming platform portfolio. So that's not going away. Now it's being more enhanced. Same with the OTT side. It's still there and it's going to be more enhanced. That's the biggest thing. Watson is it has its own life. It's its own living thing, but it's an enhancement to the OTT side and an enhancement to the streaming side. You can go on both sides.
Tim Siglin: As a solutions engineer, what kind of challenges would you be facing that a customer has that they need to use Watson to do that would be sort of out of the ordinary?
Scott Grizzle: Let's say you have a major pharmaceutical company who is doing their town halls. So corporate communications, enterprise communications, they still need to do transcripts. They need to have closed captioning on the fly or on VODs for compliance things. Maybe they need to do translation and things like that. So that's one way to go out. Or let's say you're working with a movie theater. They've got to do the same thing. You want to give them all the metadata information so they can do better trailers, more marketing information and SEO.
Tim Siglin: What's the feedback loop to bring back into Watson to understand that this was successful in terms of creating a movie trailer that sold more tickets or got more people? How does that kind of real-world interaction back to Watson work?
Scott Grizzle: The biggest thing is, it's still people. So, the whole thing with Watson and the Morgan Trailer--the team out there did it, it edited the video but then they came back to it like myself or my colleagues and we say, "Hey, we need to have these features added" and we'd go through or "We need to get this train to do this functionality." The whole thing with Watson is you have an idea and you can train it to do more and expand.
Tim Siglin: Is artificial intelligence here to stay in the video world, replacing humans doing metadata entry manually?
Scott Grizzle: Ironically, I have a friend who works in the Pentagon. He actually does a lot of video for them. He cannot hire staff. He can hire equipment. He can do services.
Tim Siglin: I actually worked in the DOD and I understand that completely. It's different colors of money or different pots of money.
Scott Grizzle: Yeah. So that whole thing is with that, imagine you can have all this longform editing that has to come through. So you need Watson or something to go through and cut out all the bad video and give you the highlights. Then you go through and you can go from there. You can segment. You can do scene selection, things like that. The scene selection is pretty cool. We actually ran a couple commercials through this week and it broke it down for every shot, where it sees a theme in the picture and the theme inside of the text. So it'll go through. Well this scene still falls under this scenario, but the next one. Here actually starts the next overall scene.
Tim Siglin: To a certain extent, there's some production creative decisioning where you might want to do video follow audio. Where you hear the narration first and then you cut for effect. It would also have to understand those kinds of scenarios where you're intentionally overlapping. Interesting. So what, you've been with IBM now for, what, a year?
Scott Grizzle: Almost two years. It'll be two years in January.
Tim Siglin: What's sort of surprised you about being part of a large company like IBM?
Scott Grizzle: Well the thing is, the big part of IBM is obviously the resources they have. They're such a massive company. But at the same time is, we're still very agile, very nimble. So allowing us to build the products we need to build with our teams and go that route. It's not like everything is built through bureaucracy. It's still the same team that was building before. It's the same team building it now. Just now there's more resources and if we have an issue, we can reach out to another unit who can help us out.
Tim Siglin: Very good. What else are you looking for at the show?
Scott Grizzle: You know I just come to the show for networking. Coming here for me I think 18 years, I started back in '97, when, technically, I was more amateurish. The whole thing is, you see the new technology, you see people you've known for those years and that's why I come to the shows is to catch up, to network, to see what projects they're working on and see how I can help them and how they can help me. That kind of thing. That's really what the show is about.
Tim Siglin: Good deal. Scott, as always, enjoyed it.
Scott Grizzle: Thank you.
Soccer fans can view World Clips soon after they occur, then share them with friends, thanks to an AI-powered system that speeds the workflow.
Watson Captioning will learn over time, becoming more accurate as users make edits and add their own custom glossaries.
AI, machine learning, and neural networks are no longer the sci-fi dreams of tomorrow. They're here today, and IBM Watson is leading the way. Learn more about how it works in our webinar Thursday, December 14.AI
Streaming Media's Tim Siglin interviews VideoRx's Robert Reinhardt at Streaming Media West 2017.
Streaming Media's Tim Siglin interviews Microsoft Principal SDE, Communications & Media Andy Beach at Streaming Media West 2017.
Companies and Suppliers Mentioned