While IBM Watson has risen through the ranks of Big Blue to become a household name, little has been written about the IBM Watson Media arm which was quietly formed within the supercomputer behemoth five years ago – before officially launching in 2016. Faultline Online Reporter grabbed some time with the relatively new division at NAB to discuss broadcaster demand trends as well as its direct work with the FCC – as IBM Watson Media competes more and more with Microsoft in the media space.
Watson Media covers a number of verticals, from fields perhaps perceived by the mainstream as ‘uncool’ such as captioning and archive data tagging, to more buzzwordy areas like AI and esports. The company divides its products across five core technologies – Captioning, Video Enrichment, Recommendations, Highlights and Streaming. IBM’s cloud infrastructure clout naturally helps sell these services, based on the important latter streaming segment which handles video hosting and transcoding as well as speech to text and analytics in one cloud video platform. Of course, these technologies and services are all exclusive to IBM Cloud.
But the captioning element, according to Senior Offering Manager David Kulczar, has only been built over the last few months. We mentioned this was interesting considering we have spoken to a number of smaller vendors offering clever captioning techniques within the past year, some of which have made a point of knocking bigger rivals. These vendors have said that while Watson and Google might do a stellar job of transcribing content using automatic speech recognition (ASR) technologies, they apparently lack full suites of localization and packaging technologies.
Of course, it’s easy for a niche specialist to poke holes in a goliath like IBM, so Kulczar simply said IBM sells tools to niche players with overlapping customers, so the negatives highlighted by smaller rivals should not necessarily be perceived as a negative.
Watson Media addresses the minefield of captioning live events, as well as on-demand content, by promising customers continuous improvement of the Watson AI over time, as well as using existing Newsroom Control System scripts to pre-train Watson, to ultimately deliver highly accurate cloud captions. It also claims to have more direct speech to text APIs than Microsoft and incorporates features such as custom corpora (large, structured set of texts) and custom vocabulary to enhance accuracy of first-run caption scripts. One particularly treacherous mine comes in the form of hefty fines issued by the FCC for captioning errors and Watson Media says working more directly with the US government agency is crucial to helping broadcasters avoid needless fines.
As for customers, Watson Media only really slid into the public eye following its deployment at the US Open tennis tournament in 2017, applying a suite of technologies for analyzing video, images, language, sentiment and tone. Watson Media allows content owners and distributors to understand the unstructured data behind content on an “unprecedented level” – giving advanced video insights for decision making. The United States Tennis Association (USTA) has employed the technology to automatically select and produce match highlights, and Sinclair Broadcast Group also tapped Watson Media last year to automate its live broadcast closed captions.
Watson Media has since gone on to power Wimbledon and the Masters, for example producing 3 minute round highlight reels applying AI to detect aspects like reactions from crowds (noises, celebrations etc.), which are automatically spun into a highlight reel. We pressed that there must surely be a human barrier as a final port of call before being broadcast, but in the high stakes sports game, getting these highlights out in seconds is more important and clearly Watson Media is capable of producing error-free reels.
Kulczar whipped up an example demo from Wimbledon showing how highlights are automatically given a rating from 0 to 1 based on the significance of an event. Applying AI and automation to content which has so many subjective metrics, in such an objective manner, really impressed us.
That said, normally our stance towards AI and machine learning is cautious to say the least but faced with a beast like Watson, doubting its credibility is nigh impossible. However, we feel the Watson Media team was deprived of flaunting the extent of Watson’s full AI capabilities at IBM’s NAB booth in terms of demos – an observation which is sure to change as the fledgling unit finds its feet on trade show floors in the coming years.
IBM Watson had no announcements to share at NAB but Kulczar highlighted the company is currently doing a lot of work in the esports space, so we can expect some news in the next couple of months.