Podcasts are audio files that can contain literally anything. Typically, podcasts are usually produced in some type of talk show format with one or more persons hosting the program. Topics available are as varied as the human experience. In addition to amateur endeavors, professional broadcasters have found the podcast to be an effective venue in which to distribute their content where it can be listened to at any time and on a global basis. Most radio talk shows including mine for example, are now available on the web as podcasts. Listeners can subscribe to the podcast which lets their computer automatically download and save it into an attached media player such as an Apple iPod. But even after you discover a podcast that you like, you still really don't know what's on it until you listen to the entire download. Oh sure the podcast may have some associated text along with it but it's minimal at best. And even if you know what's on it, the only way to find it is to try and intermittently scan by fast forwarding a few moments to see what is being said. It's clumsy at best.

But now a new website has come up with a way to let you literally pinpoint what you want to hear within the podcast itself. To do this, it uses speech-to-text recognition.

PODZINGER is an amazing new website that literally lets you search for any spoken words within an audio podcast. Until now, most podcast searches worked by searching the limited description text such as Subject or Category or the small amount of metadata that had to be included manually by the podcast creator. PODZINGER works by literally listening to the entire podcast and creating a text transcription using a sophisticated speech-to-text process. Once the text file is created, the PODZINGER website lets you search the file for any number of search words and phrases.

For my example, I told it to look for the words "Computer America" which also happens to be the name of my syndicated radio show. PODZINGER began a search of its over 200,000 podcasts and immediately found around 80 podcasts in which those exact words were spoken. Most were my show podcasts but it actually found a few others that just happened to have someone talking about my show on some other program! Needless to say that if it weren't for PODZINGER, I might never have known about it.

Once all of the podcasts are located, PODZINGER presents you with a chronological listing of each one along with the name of the actual podcast and any description of it that's made available. But then it gets better.

Each podcast is further broken down to show the actual sentence that contains the search words along with the actual running time where it was spoken. Amazing. To hear it, you can begin playing the podcast and then literally zoom in to the exact moment where the search word was spoken by clicking on the time segment at the beginning of the sentence. You can see the time of the playing podcast advance to the given moment and then you hear what you are reading. It's uncanny how accurate this can be.

Oh sure, granted that speech-to-text recognition isn't 100 percent accurate and some of the transcriptions can be a little bizarre, but it's really easy to deduce what is actually being said. The bottom line is that you can find whatever spoken word you want right down to the second it's being spoken.

This is an incredibly useful tool for anyone who needs to locate something of importance within any podcast.

PODZINGER continues to increase the podcasts they scan and convert. If you wish to include your podcast in their search, PODZINGER lets you register your podcast address along with the HTML code to insert. Or if you want to add iTunes and Yahoo Podcast addresses, they will add those as well.

PODZINGER will notify you when your podcast is ready to be "ZING'd."

Currently PODZINGER works for English and Spanish, and is a free service.

I always enjoy it when I see something new that was created from a clever combination of existing technologies. That's what PODZINGER has done in this case. None of the individual components used are brand new but their clever combination has resulted in a podcast-content searching method that until now was just not possible to do.