nScreenMedia OTT multiscreen media analysis

NAB 2019: Three ways AI is impacting video today

AI artificial intelligence splash

AI may not be the headliner it was last year at NAB but the technology is making the transition from hype to happening. Here are three real applications of the technology I heard about at NAB.

David Mowrey IBM Watson

David Mowrey, IBM Watson Media

For me, NAB 2019 opened with artificial intelligence sharply in focus. I sat down with Dave Mowrey, VP of Product and Business Development, IBM Watson Media and Cloud Video, to discuss how AI is used in media businesses today. Mr. Mowrey reminded me that:

“AI is being used behind the scenes every day in real ways that aren’t always obvious.”

His comment resonated in many conversations I had with vendors on the show the floor. Here are three practical applications of AI and machine learning that I heard about in the last couple of days at the show.

Taking a bite out of churn

SVOD services are working hard to build sustainable businesses online. One of the most obvious ways to do that is to hold on to a hard-won customer for as long as possible. Paywizard is using its AI-powered Singula business iBhavesh Vaghela Paywizardntelligence platform to help video providers do just that. Bhavesh Vaghela, the company’s CEO, told me Singula was being used by one established UK SVOD service to monitor free trial conversions.[1] Singula was able to predict with 95% accuracy which customers would churn before the end of a free trial.

Leveraging the churn information, Singula was able to make a big impact on free-trial conversion performance. The SVOD provider was already using webchat to talk with customers when they were in the process of canceling a free trial. This “reactive” approach convinced 22% to stick with the service. Leveraging the cancellation prediction to “proactively” webchat with likely cancelers, Singula boosted retention to 44%.

Boosting metadata detail

One of the key areas Mr. Mowrey identified where AI could help is in liberating “dark data” from video. Fox Sports used IBM Watson Media to process 58 years-worth of world cup action to identify clips of key action segments. Fox used the clips in a customer feature on the website that allowed the viewer to watch classic moments for favorite teams and players. The production team also used the clips during match broadcasts.

Verizon Digital Media Services was demonstrating its partnership with Microsoft Azure at its booth on the show floor. As part of the relationship, VDMS is using Microsoft’s Video Indexer to help liberate “dark data” from videos it is managing. The Video Indexer identifies things like scenes, actors, and company logos in the video. It creates a time index to all occurrences of these objects in the video. The Indexer also parses the closed captioning text to enhance metadata detail about the video.

Optimizing the encode

Today, when video is digitized using an encoder, a single profile is used to optimize the results. For example, Top Gun might be encoded using an action movie profile. The profile requires a very high bitrate for the digital version to ensure fast action scenes are rendered with great fidelity. However, many scenes in the movie contain no fast action. These scenes could be encoded at a much lower bitrate with no loss in quality. The resulting digitized file or video stream of the movie is much bigger than it needs to be.

Bitmovin is using machine learning to optimize the encoding of movies based on individual scenes rather than a single profile. The encoder can recognize an action scene and use a higher bitrate than a non-action scene. The approach results in a file or video stream that is much smaller than the single profile approach. Smaller and lower bitrate video files save video providers money by cutting the cost of video storage and streaming bandwidth.

To find other ways AI is impacting media, point your browser at www.AIinMedia.com.

Why it matters

AI is beginning to reap real benefits for video providers.

It is being employed in real applications to:

  • Reduced churn and boost free-trial conversions
  • Improve metadata richness and depth
  • Save video providers money for video storage and streaming

[1] Free trial conversions – the rate at which people using an SVOD service on a free trial convert to a paying customer


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.