hey all, so i've been thinking about this for a while. say you're in a crowd of a hundred people, the human mind can channel out 99 others and "listen" to just one. To mimic this, they have directional microphones. An interesting problem is to come up with some algorithm that can channel out (and actually recognize the speech, to some level of accuracy) a hundred speakers with a single microphone. I haven't read too much about it, but please let me know what you guys think of it. For those who want to read more about it, this idea goes by the name of "speech diarization" in the signal processing literature.
Subscribe to:
Post Comments (Atom)
Posts
3 comments:
How close is this to multi-touch screens? This is a great get rich quick idea haha, I like it!
Can people turn what's unique about each voice into electric signals? I think that might be the hardest problem.
But i suppose they can measure out the intervals and what not...
It'll be pretty cool though, next James Bond gadget! :D
i remember my math teacher telling me that if a person stands on each focal point in the oval office... like no matter how loud people are in the room... if one person whispers on the focal point the other person can hear it loud and clear... maybe you could apply something like that?
Post a Comment