Just read that Google has launched an audio indexing service into it Labs. Dubbed ‘GAudi‘ (shocking name, though I can’t talk) it is designed to work with YouTube to analyse every word said in a given video clip. The transcript will then be added to a database for searching much in the same way as you would search for text-based web content.
Launched a couple of months ago, it received such a great response that they’ve formalised the project which now includes searching within a video and a ‘more robust interface’. They say that it is in beta at the moment and is only indexing political sources.
I see another use for this. Surely this would be a fantastic tool to increase accessibility of video content online? Can you imagine installing a browser plugin which, every time a video appeared on a website a user was visiting, displayed a small button to ‘turn on Google subtitles’. It could work like the UK’s 888 Teletext subtitling service. It would be invaluable for those hard of hearing. Though I would hope the accuracy is improved from the Teletext ‘bored student in basement who can’t spell’ effort.
It could be extended as on online app. Upload your video, Google ‘listens’ to it and outputs the text (as XML maybe) for you to create your own subtitled version. I can think of a few projects I’ve been involved in over the last couple of months where this would have cut the manual overhead drastically. In fact, throw in a bit of Google Translate for good measure and suddenly a whole new world of content opens up.