Apple’s Siri is famous for her human-like voice, but in 2017 she got a gender tweak. “Apple released the latest iOS update in 2017 and customized Siri’s pronunciation to be gender neutral,” writes Rachel Hammill on AppAdvice. “She has been available as Siri on iPhone 4S, iPhone 5S, iPhone 5C, iPhone 6, and iPhone 6S.”
The “gender neutral” take may be sweet music to Siri’s ears, but the gesture also means the world to the scientist who developed the voice. Maya Wright from California joined scientists in the late 1980s to try to improve on IBM’s successful speech synthesis program Deep Blue. Although Deep Blue was identified as being capable of being a “computer,” Wright says that it was more like an artificial intelligence that could learn to perform a task. Siri learns by listening to thousands of voices, says Wright, and the best generalizations of sound that become nodes in a neural network are imitated.
Wright explains that as every voice is copied and pasted, they become nodes in a neural network. Each time the nodes are recorded, they also gain knowledge about how to make Siri more useful by hearing thousands of comments on the thousands of accounts. This means Siri is comprised of hundreds of thousands of generic voices. “If a computer could be so efficient at finding something in your history, why can’t AI or computers?” Wright asked. “Why does it have to be so hard for us to make robots that communicate?”
Read more about Wright at Smithsonian, or watch the Smithsonian Channel’s documentary clip below:
The 15th annual Innovations in Mobile will take place this Friday at the Sheraton Philadelphia Hotel and Towers. WGN America will air the three-part series, “Maze Runner: The Death Cure,” on Friday.