Take the world’s largest collection of online videos and add speech recognition to aid searching, and what do you have? Google Audio Indexing, nicknamed “GAudi”, perhaps the next step in taking the “gr” and “ail” out of the Holy Grail of mass public automatic speech-to-text. Aside from its mainstream value for media research, we’re one step closer to automatic captioning. Without such tools we will never be able to have captions for the millions of low-volume videos being posted daily.
Focused right now on videos pertaining to the upcoming 2008 US elections, Google has promoted this speech recognition gadget onto a page of its own on Google Labs.