Online services are made and unmade by how well they listen to feedback. Keeping track of what’s written is easy enough, thanks to blog search services like Technorati (which is much faster these days – thanks, guys). Keeping track of what’s spoken is significantly harder. Up until recently, if feedback came in podcast form, I had to be made aware of it through a written source first, which meant I risked missing a lot.
I’m therefore quite thankful for AP Tech Writer Brian Bergstein’s “Podcasts converted to text” product review, which pointed me at Blinkx, Podzinger, and Podscope. (Side note: Since articles from AP are heavily ‘reprinted’ online, why wouldn’t the AP simply distribute an HTML version with links to the relevant URLs built in?) Bergstein likes Podzinger for the transcript snippets but I find the replay option utterly useless (requires RealPlayer & IE 5.0 and higher) so Podscope gets the nod. It’s probably not statistically significant, but for the relatively narrow set of queries I searched for Podcast also returned slightly more results. Since each service’s search is still imperfect and limited to a subset of podcasts I’ll be adding alert feeds from both Podzinger and Podscope to the ego and feedback sections of my RSS reader. Someone say my name in an indexed podcast, so I can see what the algorithms make of ‘Yardley’.
Rumors that Google’s buying face-recognition software company Riya make me wonder if voice-to-text companies will also be snapped up by Google and others looking to compete with Google. Any automated way of taking non-searchable content and making it searchable seems like a natural fit – if the technology hasn’t been developed in-house already.
