Thursday, February 7, 2008

Finding Podcasts

The most interesting search engine I checked out was a thing called "Podscope". It claims to be able to search podcasts, both audio and video, for the actual text of spoken words.
My first attempt was to search for "Telecaster" which is a type of electric guitar. No hits. I got less specific and used guitar, which immediately brought up a bunch of podcasts, including NPR music reviews.

Next, I tried "Donahue" (no, not "That Girl's" squeeze, but somebody who plays a Telecaster). Here is where things got interesting. The first couple of hits were for some radio show called "Dawn and Drew". So it appears that Podcast is taking the actual phonetic "noiseprint" waveforms of the text you type in, and comparing that to actual waveforms of podcasts. I can imaging that "Welcome to the the Dawn and Drew Morning Show" would look a lot like "Welcome to the Donahue Mourning Show", when he finally admits to himself that his wife was right-on about the consequences of supporting Ralph Nader. But that's water-under-the-bridge.

This process reminded me of the gymnastics Jin has to go through to type written Chinese into the computer. First, a program sits in-between any other program she is trying to use. You type into the program the romanized phonetic equivalent of the Chinese word you are trying to type. A list of phonetic matches comes up, and she chooses the most appropriate characters from a list of 30 or more possible matches. If you have found what you are looking for, you paste the characters into the document you are writing. Replace drawn Chinese characters with waveform "characters" and the Podscope process is about the same - but without the element of human selection, which accounts for so many bad hits in its result list! But since they give you the whole list, you can make that selection. This is ingenious! I can't wait to see how much it improves...

1 comment:

Ramona said...

Wow, this Podscope tool is fascinating, and Jin's use of technology for Chinese characters is also fascinating. Thanks for sharing this information!--Ramona