Podcast Production Services features a wide range of services. Everything you need to create and deliver your Podcast.

Thursday, September 14, 2006

Surfing the Internet for Spoken Words

New Technology Allows Searchers to Scour
Online Audio, Video to Target Advertising
By WILLIAM M. BULKELEY
September 14, 2006; Page B3

One of the charms of Internet video and audio is that Web sites featuring such offerings are largely free of the advertising cluttering television and radio.

That may be about to change.

Several small companies are starting to pitch advertising links using their software that will search every word spoken in Web-borne video soundtracks or Internet audio programs known as podcasts. The new technology, from companies including Podzinger Inc., TVEyes Inc. and Blinkx Inc., uses voice-recognition software to translate spoken words into text or audio-wave forms that can then be searched.

Identifying spoken content of audio and video clips results in more-relevant results when using a search engine to look for a particular item or topic. From there it is only a short step to also use the new technology to match related advertising with the search results -- much as Google Inc. and others do for searches of text-based material.

Major search portals, including Google, already offer searches for videocasts and audiocasts. But they search for text "tags" -- a few words of summary created by the producers of the content that may not fully describe all the content of the audio or video material.

The new technology makes it possible to take a searcher directly to the portion of a podcast or video where the speaker discusses specific topics of interest, such as mutual funds, cholesterol or Lindsay Lohan. On the side, the search page can display ads supplied by Google or Yahoo Inc. based on the search term, with the site that serves as the host for the search getting a cut of the ad revenue.

Podzinger, of Cambridge, Mass., which provides audio search on its site and for some partners, says the ability to find words in videos fills a huge gap. "Audio and video have been a black space that cannot be discovered by traditional search engines," says Alex Laats, Podzinger's chief executive.

Also, the traditional "tag" searches typically take the searcher to the beginning of what may be a very long audio or video interview, for example, without telling the searcher how to quickly hear or see what they want.

Software products from TVEyes take the Web surfer directly to the place in the video where the search word is spoken in podcasts available at Evoca.com, a podcast-hosting site based in Savannah, Ga. David Ives, president of TVEyes, Fairfield, Conn., says his company's PodScope software also will analyze advertising clips for key words that are relevant to a user's search so ads can be matched with search requests. He says Time Warner Inc.'s AOL is testing PodScope search podcasts.

With the spread of video and audio on the Internet, "The ability to target advertising to content is a major leap forward," says Allen Weiner, an analyst with Gartner Inc. He says it may spur "monetization of video."

Blinkx, of San Francisco, provides search technology to sites like FoxNews.com and Lycos.com. It also sees opportunities to sell search-related advertising for audio and video content related to travel and personal finance.

The new search technology captures only part of the Web's audio content. Podzinger says it is indexing for search some 300,000 regular podcasts, or 30% to 60% of the estimated 500,000 to one million podcasts available on the Internet.

Even the best speech-recognition technology has trouble understanding many speakers. People with accents or colds confuse it. Music in the background causes trouble. Suranga Chandratillake, founder and chief technology officer of Blinkx, says accuracy ranges from 60% on amateur videos to close to perfect for trained newscasters in professional studios.

Searches often return many irrelevant videos. For example, looking for "online investing" on Lycos, which uses Blinkx software, gets 19 results including, logically enough, an interview with a low-priced stock-trading firm but also, inexplicably, a BBC-TV clip about a nurse murdering elderly patients.

Still, the speech-recognition technology picks up many words and takes the searcher directly to the relevant portion of the recording. For example, if a Boston sports fan wants to find out if outspoken Red Sox pitcher Curt Schilling has ever opined on New England Patriots quarterback Tom Brady, he can search the Internet archives of sports-talk station WEEI for "Schilling Brady" and find a link. Not surprisingly, Mr. Schilling thinks Mr. Brady is great: "I'm a huge Patriot fan for a lot of character reasons. Is there any doubt that Tom Brady is going to make the four guys he throws to good?" Mr. Schilling asks.

Bill Alfano, director of marketing for Entercom Inc., based in Bala Cynwyd, Pa., says his company started using Podzinger last month at its Boston sports-talk station WEEI because "we were blown away by the technology."

He says it provides a way to let the 100,000 WEEI listeners who have registered at its Web site retrieve segments they might have missed. The station hasn't started to link ads on its Web site to the audio search capability, but Mr. Alfano thinks there is potential. "If Tom Brady says, 'I love to go to the Pro Bowl,' at the end we could run an ad for a trip to the Pro Bowl. People almost expect that," he says.

Mr. Ives of TVEyes says that on sites it has indexed, PodScope continuously looks ahead 30 seconds as a viewer watches a video. Then "it puts a contextually relevant clickable ad near the organic content," without interrupting the video the way a traditional video ad would. He says that "in early tests we find the click-through rates are a multiple" of clicks on random ads.

One believer in the technology is Leo LaPorte, who hosts radio shows and podcasts about technology. Mr. LaPorte, who produces 50 hours of programming a month, plans to start using Podzinger on his sites next week.

With the technology, he says, listeners who want to hear again what he said during a program about Fujifilm's Finepix digital camera, for example, can search and find it. Then the search engine can provide click-through ads from half a dozen photo stores with prices for the camera. "It's considered a service" by searchers, he says, and he can get $5 to $25 each time listeners click on ads after such searches.

Mr. LaPorte says he is hopeful that audio search will boost his revenue from "a couple of hundred thousand a year now to over $1 million a year. For a guy working in an attic, it's a viable business."