Blinkx Redesigns

Meghan has sent me this email about the re-design of blinkx. As per the email:

blinkx, the world’s largest video search engine, redesigned their website to allow people to unleash their “inner couch potato side” with new Inform Me, Entertain Me, and Give Me Your Own Channel buttons on the Website.  Viewers can now choose which types of videos they want to watch, then sit back and let blinkx show them what they asked for.  From the latest headline news stories about David Beckham’s wish to remain in Milan to funny skateboarding dog videos, blinkx allows you to discover content and sit back, relax and enjoy.  You no longer have to know what you’re looking for; blinkx will just give it to you, similar to your TV experience.  If you want more, you can click on more information to deep dive into the videos you really care about through a tag cloud, facial recognition and speech cues.

Their press release has these following features:

- Speech tags: leveraging patented speech and concept recognition software, blinkx technology listens to the speech track of a given video, identifying and extracting speech tags that represent key words and phrases spoken within the video. These tags complement text metadata tags, improving the description of the video and lighting up as the corresponding words are spoken in the video. The speech tags can be used to navigate – both within the video, allowing users to jump directly to the point in a video that a word is spoken, and also from the video, as a starting point to find other relevant videos. For example, a video about President Barack Obama’s inauguration might include speech tags such as “Barack Obama,” “oath” and “Lincoln Bible”. If a user were interested in learning more about the history of the Lincoln Bible, he could click on the speech tag to retrieve and explore more videos about the subject.

- Facial recognition: using its unique visual analysis capabilities, blinkx automatically identifies and catalogs recognizable faces within a video. In the example of a clip of the inauguration of President Barack Obama, users would be presented with thumbnails of the different faces that appear in the video, including the President, the former President and the Chief Justice of the United States. If interested in a particular person in the video, users can simply click on the relevant face to jump directly to the scene that features that individual.

- Scene change analysis: by watching each video frame-by-frame, blinkx can detect scene changes and match videos that share similar visual components. For example, news videos on the presidential inauguration might come from different television networks, but if they share similar scenes from the event, then blinkx presents the videos as a set of related results. In addition, the scenes of each video are shown as a storyboard, allowing a user to track, visually, where they are within a video – these scenes can also be used as chapter anchor points, by clicking on one, the user is taken directly to that scene.