beta

A lot of talk today on Podcast searching – see Steve and Shel’s links

I had meant to post something today on this topic – was listening to Neville and Shel earlier

Disclosure – I’ve quite a bit of experience in the audio search space, so i know a little bit

Some facts - no product is 100% accurate - the better ones are phonetic based, and use Hidden Markov Modelling (HMM) - these products can be trained to be better – costly exercise to get right - this technology is expensive - work reasonably ok with “BBC type voices” - the best one is Softsound - Softsound is owned by Autonomy, which in turn is involved (see press release) in Blinx

Also check out some cheaper alternatives - Sphinx – open source - HP Labs SpeechBot

BTWresearch, tools – for free.