YouTube has developed technology to automatically transcribe speech from a video’s audio track, and then embed that text into the video.
The process has been tested on speeches by John McCain and Barack Obama, and allows users to search for key words (such as “Iraq” or “free trade”) which will then bring up the exact position in each video where that phrase is spoken. The transcript is then available in the timeline underneath the video.
I’m thinking the service is going to work much better on “sensible” videos, as I dread to think what the automated spiders would make of karaoke.
Here’s a video of Steve Grove, Head of News and Politics at YouTube, describing the new service. Given that 13 hours worth of video is uploaded every minute, we definitely need a decent system for searching it.