Watch all red carpet interviews from Streaming Media West in San Jose.
At the Streaming Media West conference in San Jose last week I talked to a lot of companies that are helping to make video more discoverable, whether it is online, on your set top box, or mobile device.
I also had a chance to catch up with Tim Siglin, who recently wrote an article about video metadata for Streaming Media Magazine.
Tim, who feels that metadata doesn’t get enough love and attention in the industry, was surprised that I read his article from start to finish. So in our effort to sex up the topic a bit, he offered a few compelling reasons why you should care about video metadata and start to pay attention to the companies that develop metadata solutions. The conversation got steamy.
First off, what kind of metadata is possible beyond the basics of titling a video, writing a short description and tagging it with a few key words to increase the likelihood that it will be found by others?
The answer is a lot. And it starts during production.
Tim points out that as you go through the production process there is metadata on the camera. Some of that information transfers to the editing system timeline, such as Final Cut Pro, and some of the metadata generated during editing gets transferred to the encoded video. But most metadata gets lost along the production process so that the final video that is uploaded to your platform of choice has very limited accompanying metadata.
The more relevant metadata there is associated with an online video, the easier it is for end users, whether they are colleagues, customers, or consumers, to find the video and, more precisely, specific scenes within the video.
There are a few ways that metadata can be added to a video. Metadata is automatically generated throughout the production process, but rarely passed from process to process. Metadata is often added manually by the content creator when he uploads a video. Some video platforms allow end users to add metadata to a video timeline. People can add tags and comments while watching the video. This is different than leaving a comment in the comments section below the video as that information often does not travel with the video.
There are tools popping up that can take video as an input and run the video through speech to text or facial / scene recognition programs. The programs can tell you not only what words are said at what point in the video, but that Brad Pitt appears on screen at minute three or that a car appears in minute eight with a mountain background rather than a cityscape.
Hollywood studios are building internal metadata capabilities so employees can find the content they are looking for. Studios need metadata to find specific video assets within their enormous content repositories. They see the value in developing improved metadata for workflow, but haven’t exposed metadata to the general public…yet.
Imagine if you could search MGMs library for every scene from every James Bond movie that has a car chase.
Companies will probably want to restrict access to certain metadata. So they will need solutions that make some metadata publicly available but other metadata only exposed to internal employees.
Corporate training and college lectures are another market that would benefit from metadata. The content, besides being really long, is chock full of juicy keywords that probably aren’t manually tagged by the person uploading the lecture video after class ends. Imagine back in your college days searching for a physics term and seeing every clip from every class where the term is mentioned. It’s like when you search a PDF for a phrase and it highlights the phrase in yellow on every page it’s mentioned and gives you quick links to each mention of the phrase.
Tim’s most compelling reason for people to care about deep layers of metadata? Simply, the more discoverable video becomes, the more it can be monetized.
In a conversation with Paul Scanlan of MobiTV, he told me how MobiTV is working with business video content providers such as CNBC and Fox Business to make it so that a consumer can type in MSFT and discover every segment from every show that mentions Microsoft. Instead of having to watch the entire 30 or 60 minute program, the user can just watch from 30 seconds before Microsoft is mentioned to 30 seconds after the mention. Would investors pay for that? I’m guessing some would.
Or imagine a fantasy football owner trying to decide whether to play Tom Brady or Philip Rivers on a Sunday who can then search for Tom Brady and watch every ESPN, FOX, CBS, and NBC video clip that mentions Tom Brady. If you’re trying to win your league, and there’s money on the line, would you pay $5 for this type of powerful search tool? I would.
