Futurewatch: Media & Entertainment—The New TV-Web-Mobile Mashup

While Azuki is building its technologies as a service for an end-to-end media delivery platform for mobile, there is no inherent limitation to implementing this mashup concept for broadband, according to John Tremblay, Azuki’s vice president of marketing. Tremblay points out that in mobile there is an economic rationale to reduce the amount of data that is streamed, and a nonlinear approach makes that possible while enhancing the user experience.

In addition to Azuki, there are a few other companies that are helping address similar challenges for the web. Notably, Azuki’s solution relies on metadata and indexing. Interestingly, the other companies that have also drawn my attention present themselves as metadata and indexing companies for the most part. Their selling points today are around search and clip creation, which are important functions that everyone understands. However, as we look at nonlinearity in video, the possibilities such technologies suggest for the middle-ground area become intriguing.

Gotuit and Digitalsmiths have been working on this problem since before the advent of online video. Not surprisingly, online video has redefined the opportunity spectrum for them. While this middle ground has yet to reach any kind of mass awareness, Gotuit’s existing customer portfolio demonstrates the range of possibilities from programmer-created search, clips, playlists, and advertising options to user-generated programming options.

EveryZing, which entered this market relatively recently in 2006, spun out of BBN’s legendary work in speech-to-text and natural-language processing. Digitalsmiths relies on speech-to-text and computer vision, and while licensing its service to other publishing platforms, it recently began offering to include a publishing platform, similar to Azuki’s end-to-end systems approach. According to Digitalsmiths’ CEO Ben Weinberger, the paired capability to mine data with a video publishing system provides an additional layer of intelligence for monetizing video for publishers. While all these participants are approaching the opportunity space slightly differently, one of the major distinctions is whether such functionality can be automated. Or is video inherently a medium that is too complex for machine interpretation? Gotuit’s solution relies on manual authoring of metadata. According to Gotuit’s CEO Mark Pascarella, manually authored metadata allows additional creativity to be applied to an already-creative process of video production. Tom Wilde, CEO of EveryZing, considers his company’s speech-to-text and natural-language processing technologies sufficiently advanced from where such technologies started, and those technologies can process a large quantity of video efficiently and accurately. Digitalsmiths’ Weinberger agrees there’s a place for each, and there’s also a place for combining manual and automated approaches.

Also taking the manual approach is Veotag, which is finding growth in enterprise and government segments on account of implementing essential functions with simplicity for the operator. According to Jeff Paul, Veotag’s vice president of sales and alliances, as a result of chapterization of the customers’ videos, click-through rates have grown from 3% to 25% in some cases, and unique viewers have increased by 500%. These are attributed to indexing but also to the ability for users to create custom playlists and share clips easily.

Digitalsmiths and EveryZing, similar to Azuki, are relying on degrees of automation to parse video into more discrete segments for the purposes of search, clips, playlists, and recommendations. While Azuki does this along discrete time segments (in addition to using existing markers), the other companies use things such as speech-to-text, computer vision, and other processing automation to create metadata that can then be used to create clips, seek points, and more, including re-creating a timeline.

Adobe has similarly added speech-to-text conversion in its Creative Suite 4 applications, potentially changing the rules for adoption and implementation of such applications that the web is ideally suited to. Undoubtedly, there are differences in levels of technical advancements and performance among the aforementioned companies’ solutions, and this article is by no means recommending one approach over another or intending to draw performance comparisons.

What seems apparent though is that momentum is gathering behind nonlinearity of web video. How different metadata and indexing schemes play out in the industry is to be seen. There is also a dependence on different players (clients) to support the implemented features, creating another set of dynamics and raising the question again—how many players can coexist? However, none of these are new questions, and if the past is any indication, such concerns have not stopped companies from innovating and customers from adopting new internet technologies. Try counting the number of different video formats on the internet, and let me know when you figure it out.

In some cases, value propositions of these companies have not been articulated beyond the obvious search and clip creation and the promise of accompanying advertising. But it is not a stretch to envision that, given the pace of evolution in online video, by deploying such features today, publishers will essentially futureproof their assets for the nonlinear viewing experiences that the internet is inherently suited for. Increased adoption of such solutions may well define the next phase of the web video industry.