Google Duplex is an extension of google assistant that can make phone calls to real humans just by you asking it to do so.
This technology is based on the Deep Neural Network that builds off WAVENET (The voice of duplex) technology.
WAVENET is a speech synthesis program that worked by joining very short units of sound together to create speech. This was a breakthrough in natural speech synthesis when it came out.
Duplex is another neural network build on top of wavelet the final result is an AI that can have a real conversation but with WAVENET vocal precision.
According to the google's plug, the humans and umms" and “aahs”, as seen in the video above that you hear are put in some time synthetically but actually it sometimes there to signal that the system is to processing just like a human word. the public reaction to this was on the side of shocking and horror and even anger by some.
Some people thought it was very deceitful to have an AI talked to someone over the phone without the phone without them knowing but Google has made it very clear that they are going to be transparent there will be letting the people on the other end of the phone know that they are talking to duplex.
[caption id="attachment_1104" align="aligncenter" width="1347"]
Duplex actually trained in the narrow field scheduling appointments or bookings and inquiring about a business's opening hours on holidays.
To be clear duplex cannot have general conversations but I have little doubt that the scope will more in the future.
Challenges in conversation
Google states that there were unique challenges when it came to training such a neural network. It actually difficult to AI robustly understand natural language as a reply in a realistic manner, because there if lot problems when it comes to the conversation like noise, different tone, accent, ambiguity and so on.
we people take such things for granted but it's difficult to look at this through the eyes of an AI. A team researcher trying to solve this problem.
watch the below video to find out some mind-blowing Google Duplex and human conversation.