A longstanding conspiracy is the tale of how Facebook is listening in on your conversations, but the way it is actually serving you ads is much more unsettling.
There’s a difference between a wake word and general purpose speech recognition. A simple wake word can be done in simple hardware on the device, while general purpose speech processing either requires heavy, relatively constant CPU usage, or heavy network traffic to pipe the audio to a server for processing.
I think for marketing purposes you could have a hot list of marketing terms (presumably these would be scarce so sold to high bidding companies) and match against those which would be a sort of middle ground between the general purpose processing and a single wake word.
You could do it in a cheap (in terms of energy) and sloppy way where it only needs to be correct most of the time of the time to have a net positive impact on ad targeting when reconciled with other user data.
There’s a difference between a wake word and general purpose speech recognition. A simple wake word can be done in simple hardware on the device, while general purpose speech processing either requires heavy, relatively constant CPU usage, or heavy network traffic to pipe the audio to a server for processing.
I think for marketing purposes you could have a hot list of marketing terms (presumably these would be scarce so sold to high bidding companies) and match against those which would be a sort of middle ground between the general purpose processing and a single wake word.
You could do it in a cheap (in terms of energy) and sloppy way where it only needs to be correct most of the time of the time to have a net positive impact on ad targeting when reconciled with other user data.
There’s also a third possibility most people ignore for what ever reason…
Speech-to-text and send to servers. No need for heavy CPU usage that way and don’t need to send MBs of Audio files…
With the technology we have today it’s easier than ever before… “colgate” and give you right into your face an ad for toothpaste !
No need for audio or complex processing. All new models come even with AI processor units… Haha ! What a joke !