Showing posts with label phonemes. Show all posts
Showing posts with label phonemes. Show all posts

08 February 2006

Rich Media Search

Two rather large straws in the wind: Riya - which offers face recognition in pictures with automatic tagging - and Nexidia, which claims to be "the only technology to fully leverage the actual phonemes that define human speech", creating a "highly scalable tool for audio mining and speech analytics."

Assuming these deliver on their promises, they could be very big. Imagine being able to let your PC search through thousands of photos for people - or, more interestingly, groups of people; imagine being able to upload thousands of hours of recorded conversation, and find just the spoken phrase you were looking for.

Combine this with the continuing drop in the price of hard drives - and hence the ability to store as many photos and recordings as you like - and you are getting close to a situation where you could archive all the significant visual and aural moments in your life - and find them again.