Blinkx Uses Speech Recognition and Image Analysis for Comprehensive Internet Video Searches

Most video search engines search the text of tags and descriptions written when a video is uploaded.  A company called Blinkx also uses speech recognition and analyzes images.  Through media partnerships with more than 700 news, entertainment and crowd-sourced web sites, Blinkx's index includes 35 million hours of searchable video uploaded across the Internet.  (Founded by Suranga Chandratilake in 2004, Blinkx is headquartered in San Francisco and the UK.  Learn more about Blinkx in this article from the San Francisco Chronicle.) Users can search by keyword from the homepage, or may use the advanced search to specify the video's source, the type of clip, or add additional keyword parameters to keywords.  Search results generate short previews of the clips and can be sorted by relevance or date.  Videos are extracted from partner sites so that they may be viewed on Blinkx, but will also include a link to the original location.

Videos can be e-mailed, saved to playlists, and embedded in web sites or blogs.  Users can also subscribe to an RSS feed for a specific search, making Blinkx a useful tool for tracking media appearances.

We've talked about Internet video archives on the blog before, click to read more: