Asimo has recently been upgraded to be able to hear, analyze, and respond to up to three people yelling at him. With eight microphones arranged all over him, he is able to work out the voices that are speaking by isolating it from each source. It then sends the information, in this case the sound, to the voice rec. software where it attempts to make what the person is saying. This software is actually able to pick up more than the average human ear is able to pickup which is a major plus. In a simulated situation Okuno and Nakadai, the engineers, tried using their robot in a restaurant where they used a series of complicated sentences. Although the robot could only identify 30-40% of what they said, it is still an achievement.
This robot was only made to handle rock paper scissors, but it was still able to pick up over 30% of normal speech. What an achievement. I wish I was able to make software like this and use it when I got to family parties and relatives birthdays.
This is a neat invention that I only think will get better over the years when they keep adding on to the vocabulary software. I hope they get it to the point where it could be used in party situations where you can’t hear anything. Or like one site pointed out, Wall Street. For how much yelling goes on there, you might miss a few things and have to turn to your trusty robot for the answer.
Via: arnsteinlarsen, newscientist
