Teaching computers to pick a conversation out of a noisy room


#1

Originally published at: https://boingboing.net/2018/04/13/cocktail-party-effect.html


#2

Oh good. Face recognition wasn’t invasive enough.

Note to engineers: because a thing can be done, does not imply it should be done.


#3

Don’t worry. As these AI model thingies get more mature, their ability to screen out one conversation among many will diminish, eventually to the point where they don’t frequent cocktail bars any more because it’s just pointless if any conversation at all is expected.


#4

Most human beings have this ability. I happen to know that some types of hearing issues can pretty much kill that ability. Ask me how i know that, just don’t ask me in a crowded party.

Related to that, i’d ask people to remember that privacy invasion is not the only use of some of these technologies.

Think about how useful some portable face-recognition technology could be to someone with face blindness, or the way that a portable (non-recording) version of this would make environments filled with human-voice frequency noise not be just audio chaos.


#5

What could possibly go wrong?

1

(Mostly joking. Mostly.)


#6

I know this rare neurological condition exists. I also know abusive governments exist.


#7

And from these two conditions we can deduce that rice pudding exists.
That was the point, wasn’t it?


#8

Hearing aids are clearly the biggest market for this sort of technology.


#9

This topic was automatically closed after 5 days. New replies are no longer allowed.