Harry Chen Thinks Aloud

Pluggd.com, a podcast directory web site, showed off its HearHere search software at DEMOfall 2006. This technology allows listeners to search the content of a podcast and jump to a specific part of the program that matches their search.

Users just have to type in a keyword in the search box, and HearHere will display a map to indicate where in the podcast the content is likely to match a request. It also displays the related words HearHere is using to make its matching decision. For instance, if you type “PGA” in a search of ESPN podcasts, HearHere might indicate that it’s using “golf” or “Tiger Woods” to find what you’re looking for.

Here is a demo of the HearHere technology.

HearHere reminds me of PODZINGER, a podcast search engine developed by BBN Technologies. While both HearHere and PODZINGER exploit speech recognition software to map speech to search-able texts, I wonder exactly how their “magic tricks” differs under the hood?

News about HearHear was also covered by TechCruch.

It’s clear that AOL is trying to boost its business by aggressively expand its web search and advertising market. Instead of fighting head-to-head with Google and Yahoo! in general web search, AOL attempts to break into the market with a niche application that neither companies have yet considered — educational search.

AOL’s new weapon is StudyBuddy.com. According to the BusinessWeek Online:

StudyBuddy.com relies on technology from search engine Blinkx to retrieve query text and multimedia results from a database of 350,000 online journals, encyclopedias, and other sites that have been vetted for accuracy and age-appropriateness for children. StudyBuddy’s creators see it as a virtual library that brings the best of the Web’s educational and homework resources under one roof.

An AOL executive argues that this idea will work.

“Before there was WebMD, there were hundreds of health sites on the Internet and no one destination to go to for information. Prior to StudyBuddy, there was no one space either. StudyBuddy is to be an aggregation of all the educational sites,” says June Herold, vice-president and general manager for AOL Consumer and Education Services.

Read the rest of this entry »

PODZINGER is a search engine that uses speed recognition software to transform the audio into words, making podcast content searchable by text. This engine is built on 30 years of speech recognition research developed at BBN Technologies, Cambridge Massachusetts.

I wonder if popular subjects in technorati are also popular in the podcast world? Here is a quick comparison of the results:

Date/Time: January 10, 2005, 11:31 AM -05:00

• Tag: Macworld
• Technorati: 15,676 posts
• PODZINGER: 220 posts

• Tag: Howard Stern
• Technorati: 18,006 posts
• PODZINGER: 343 posts

I guess podcast is still behind blog in terms of its user base.