With each emerging content type comes the ever-present need to help users find it. While text-based search has continued to evolve, effective tools for rich media are still nascent. For the Web's hot content-type du jour, podcasting, search tools have only just started to appear, though this search niche is poised to heat up: Forrester Research predicts that 12.3 million U.S. households will listen to podcasts by decade's end; the Diffusion Group estimates the U.S. podcast audience will be at 56 million by 2010.
Podcast search engines work in one of two ways: by providing search results based on a podcast's metadata and headlines or by converting the audio to text, which is then available for search and indexing. One company focused on helping searchers find podcasts, Podzinger, takes the latter approach. Podzinger, which came out of beta just after Christmas, lets listeners search through audio podcasts similar to the way that Google provides search results for text.
Podzinger, developed at the venerable Bolt, Beranek, and Newman ("BBN") organization in Cambridge, Massachusetts, which was founded in 1948, is the offspring of a technology that BBN developed to automate speech-to-text. Known internally at BBN as "Avoke STX," it was built for the purpose of monitoring overseas broadcasts and communications in various languages. BBN's Delta division has taken that technology and adapted it to index and search audio files, and initially to English-language podcasts.
Podzinger includes a couple of fundamental and distinctive features. Podzinger produces search results with short text extracts, or snippets, and at the specific recorded time of a given podcast file. The user can skim the text snippet or listen to the podcast segment to determine relevancy—which can result in an enormous time savings. Think of it this way: when is the last time you listened to a whole album or newscast? With Podzinger, you are able to more accurately determine the content of a podcast before you invest significant time in listening. The other few podcasting search technologies currently available—like Blinkx, Feedster, Podscope, and AOL's Podcast Central—don't yet offer this capability.
Another key Podzinger feature is that searches are not limited to the descriptive information contained in metadata or the indexes that limit a podcast search to pre-defined categories—which is both good and bad news. The good is that Podzinger converts all the audio to text and that then is searchable for indexing and reference. Said differently, you can find more podcasting material on your chosen topic. The bad is that, like so many search engines, you also find a lot of material that has nothing to do with your chosen topic.
The latest Podzinger release rolls out numerous user-friendly enhancements. Users can store their podcast searches into an iPod or other multimedia player, which then allows automatic delivery of new podcast content relating to saved search expression and using standard RSS feeds. Podzinger's browser and media player support has expanded from just IE and RealPlayer to also include Firefox, Safari, QuickTime, and Windows Media Player. The user interface has been upgraded so that search results for any source are now limited in number with additional results allowing for "click for more" access. Users can now sort search results either chronologically or by relevance. Finally, after a quick registration process, podcasters can add a Podzinger search box to their site with the ability to then search for podcasts only at that site or against the full Podzinger compiled and indexed library.