Talking to computers
08
Mar
Researchers at the University of Southern California (USC) researchers are developing
a high-speed language processing system, which attempt to reproduce human efficiency of decision-making in slang, words shortening and other language vices that are embedded in everyday “While we human speakers can often understand and respond to what someone is saying to us in a fraction of a second, a typical voice interface will require much longer–often a second or two–to try to understand what you have said and respond in an appropriate way,” as said Professor David USC DeVault to NSF Foundation’s website.
a high-speed language processing system, which attempt to reproduce human efficiency of decision-making in slang, words shortening and other language vices that are embedded in everyday “While we human speakers can often understand and respond to what someone is saying to us in a fraction of a second, a typical voice interface will require much longer–often a second or two–to try to understand what you have said and respond in an appropriate way,” as said Professor David USC DeVault to NSF Foundation’s website.
Researchers are looking for new techniques that can simplify man-machine conversation, allowing the system to perform all computer processing steps required in any real-time while a user is talking, the systems available for smartphones are still slow and annoying.
The researchers note that systems often fail to realize what it means that what is said and how it should respond before the user has finished speaking.
For demonstration developed an agent for high performance game called Eve, where users describe the images they see on the computer screen and the agent tries to guess what they are talking fast and as accurately as possible, that is, trying imitate humans.
The system uses speech processing algorithms “incremental” to increase the speed of understanding agent, as described DeVault to the NSF website: “These findings underscore the importance of enabling systems to not only understand what users are saying, but to do so as quickly as a human would,” said DeVault.
It is the future coming and unlike saying that automatic translation systems are not possible, this system, although embryonic, could revolutionize the translation and conversation with machines.
