Amazon previews Alexa capability that synthesizes a person’s voice from less than a minute of audio – GeekWire

LAS VEGAS — A short movie of a child asking Alexa to have his grandma browse the Wizard of Oz is generating far more than its share of excitement soon after Amazon stated that the voice was in fact a simulation.

To make it materialize, Amazon “had to find out to make a superior-good quality voice with significantly less than a minute of recording compared to hours of recording in a studio,” explained Rohit Prasad, Alexa AI senior vice president and head scientist, in the course of a keynote address Wednesday early morning at the company’s re:MARS convention in Las Vegas.

Amazon isn’t indicating when the functionality could be launched as an Alexa characteristic, or what styles of safety measures may accompany a rollout. AI-created synthesized voices have raised issues about the probable for abuse, quite possibly remaining misused to trick persons into contemplating other people experienced mentioned a little something they hadn’t.

A set of “Responsible AI” concepts produced this week by Microsoft outlines a sequence of safeguards in opposition to possible misuse of artificial voices.

Prasad hinted at the underlying specialized approach applied by Amazon for the ability, indicating that the vital was “framing the difficulty as a voice-conversion undertaking, and not a speech-generation job.”

Introducing the demo, he explained the technological innovation as a way to preserve the recollections of liked kinds, saying it is aspect of a expanding recognition of the job that Alexa can provide as a companion.

“In this companionship role, human characteristics of empathy and have an impact on are crucial for creating trust,” he stated. “These characteristics have grow to be even much more critical in these occasions of the ongoing pandemic, when so quite a few of us have misplaced another person we love. What AI can not reduce that pain of decline, it can surely make their memories final.”

Prasad called it “one of the new capabilities we’re performing on, which enables lasting individual associations.”