Google Blogoscoped


Google has plans for audio fingerprint for content relevant ads

Juha-Matti Laurio [PersonRank 10]

Monday, September 4, 2006
16 years ago4,231 views

This sounds a problem from the side of user's privacy:

"The idea is to use the existing PC microphone to listen to whatever is heard in the background, be it music, your phone going off or the TV turned down. The PC then identifies it, using fingerprinting, and then shows you relevant content..."

More at The Register;

Ionut Alex. Chitu [PersonRank 10]

16 years ago #

This is old and just an experiment.

dpneal [PersonRank 10]

16 years ago #

Peter Novig (head of research), said it will appear in Google software "sonner rather than later"

Ionut Alex. Chitu [PersonRank 10]

16 years ago #

Maybe, but they weren't able to continue providing closed captions from TV shows. This is more complicated than that.

Milly [PersonRank 10]

16 years ago #

I wonder how many ordinary surfers are aware that the Flash plugin within their browsers has the means to access (and transmit information from) their microphones and webcams? And that a third-party Flash advert network could probably make hay with this idea more easily even than Google (what with having fewer scruples, and watchdogs, and all)?

Gary Price [PersonRank 10]

16 years ago #

In terms of searching...

Several services (mostly fee-based) provide searchable content either via closed captioning or speech to text.

1) TVEyes
All major U.S. news nets and many local stations.
Some UK too.
Fee-based, real time alerts. Transcripts.
They also offer a FREE service using open web video from AP, Reuters, MSNBC. No transcripts however.

TVEyes also uses some speech to text for

2) Critical Mention

3) ShadowTV

4) Also, a new service starting soon named RedLasso.

Finally, another company that's been in this space for many years is Nexidia.
What makes them different is that they don't use captions nor speech to text in the traditional manner. They break spoken language down into phonetic sounds (phoneme) for more accurate and faster indexing. About 400 phonemes in all spoken languages. About 40 per language on avergage. Often used by call centers but expanding into other areas.

5) Blinkx and Podzinger also use speech to text.

Josue R. [PersonRank 10]

16 years ago #

maybe related to Google's search by voice, a GLab paper?

Forum home


Blog  |  Forum     more >> Archive | Feed | Google's blogs | About


This site unofficially covers Google™ and more with some rights reserved. Join our forum!