The San Francisco company, which launched its service about six months ago, uses voice-recognition software to search multimedia files, delivering what it claims are more accurate results than the major portals, such as Yahoo Inc., Microsoft Corp.'s MSN and Google Inc.
Blinkx's site focuses on TV video, which it searches and shows through partnerships with content providers, such as CNN and Fox News. In order to gain access to content not available for free over the web, Blinkx, like other video search engines, has to negotiate agreements with the content providers.
In the case of podcasts, which are programming created by individuals; or video blogs, the multimedia version of weblogs, the content is available directly on the web, and therefore searchable by Blinkx's web crawlers, Suranga Chandratillake, founder of the company, said.
Blinkx decided to start searching the new forms of web content because of demand from people using its service and the fact that the company has seen a 25-fold increase in the amount of the material available on the web, Chandratillake said. Podcasts make up 95 percent of the content, and video blogs 5 percent.
Blinkx currently tracks 20,000 podcasts over the web, representing 500 hours of audio content a day, Chandratillake said.
"We are the biggest podcast repository in the world," he said.
That may be so, but for now, the company isn't making any money. Funded by private investors, Blinkx does not charge for its services. The company plans to eventually build its business model around online advertising.
"Getting the product right, the user experience and getting content are much more important right now than making a revenue stream," Chandratillake said. "We know we can do that (eventually)."
Blinkx, however, will have to move quickly. Google, MSN, Yahoo and America Online Inc. are all moving fast into video search, with the latter debuting on Wednesday voice-recognition search on its free web portal, which launched in beta this month.
Nevertheless, Chandratillake claimed his company does video search better.
"To do pure speech recognition across content is really a very complex problem, and not one that traditional search engines know much about," Chandratillake said.