We all know that Google is doing a lot of research in language technology, and more in particular also in speech technology.

They released now some proof of what they will be able to do without any doubt later on a massive scale: speech-to-text conversion on video's. They did it only with some political video's now.

Read more on:
http://googleblog.blogspot.com/2008/07/in-their-own-words-political-videos.html

Sweet, I hope we can search WITHIN the videos in ~2 years. :)

I just wish they could help opensource voice recognition the way they're doing with OCR

Haochi, why do you think 2 years? It seems they are ready now for english spoken video's. They just have to scale it further...

"Bubble sort" doesn't return anything here.

Ricardo, more correctly I should have set [bubble sort] in italics or used square brackets... I was referring to the keyword search not a quoted phrase search.

Edit: OK, changed now and removed the quote signs around the queries.

> It seems they are ready now for english spoken video's.

Not at all (and the question is if they'll ever be). One thing is when you have a presidential candidate, speaking loud and clear to the masses, without much background noise, using a high-quality microphone, and going out of his way to be easily understandable (not to mention using proper grammar).

Totally different thing is when you have some random guy sputtering something to (and away from) a "camera" on a mobile phone, which even native speakers have trouble comprehending.