“Hey Alexa, what’s another word for ‘creepy’?”
In a demo that has drawn comparisons to dystopian series “Black Mirror,” Amazon revealed that it has developed a way for its Alexa voice assistant to replicate the speech of a dead relative — based on less than a minute of recorded audio of the original person.
The ecommerce giant showed off the new technology Wednesday at its re:MARS conference, Amazon’s global artificial-intelligence event for machine learning, automation, robotics and space. In a video demo shown at the event, a young boy says, “Alexa, can Grandma finish reading me ‘The Wizard of Oz’?” — whereupon a synthesized voice of the grandmother emanates from an Amazon Echo Dot smart speaker.
“As you saw in this experience, instead of Alexa’s voice reading the book, it’s the kid’s grandma’s voice,” Rohit Prasad, Amazon’s Alexa AI senior vice president and head scientist, told attendees.
Amazon has already developed voice-synthesizing tech to let Alexa mimic the voices of celebrities, including Shaquille O’Neal and Melissa McCarthy. But that has previously required an individual to record dozens of hours of audio. Amazon now has developed a way to replicate a voice in high quality using less than one minute of recorded speech, which the company’s engineers were able to do “by framing the problem as a voice-conversion task, and not a speech-generation task,” according to Prasad.
In introducing the demo, Prasad said that the ability for Alexa’s AI to emulate the speech of a deceased relative can represent a way to preserve the memory of a loved one who has died. “We are unquestionably living in the golden era of AI, where our dreams and science fiction are becoming a reality,” he said.
Watch the Alexa demo in the video below: