Amazon’s Alexa is nearly like a member of the family to many, with individuals more and more changing into depending on it for on a regular basis duties. Now the corporate is working to offer the machine a new function that would sound thrilling to some however eerie to others. At its annual re:MARS convention on 22 June, Amazon revealed that Alexa may quickly be capable of recreate anybody’s voice — even when they’re lifeless.
The re:MARS convention is Amazon’s annual international synthetic intelligence occasion the place the e-commerce big showcases its progress with machine studying, automation, robotics and area. The voice replication function was proven in a video demo to attendees on the convention held in Las Vegas.
Extra about Amazon Alexa’s AI voice replication tech
Demo reveals Amazon Echo Dot talking in the voice of a boy’s grandma
Rohit Prasad, Amazon’s Alexa AI senior vp and head scientist, offered the demo. In line with Prasad, Alexa will be capable of recreate the voice of a lifeless relative primarily based on lower than a minute of audio recording of the unique individual.
“Whereas AI can’t eradicate that ache of loss, it could positively make their reminiscences final,” he mentioned, earlier than the demo was proven to the attendees.
Within the demo, a younger boy asks Amazon Echo Dot, “Alexa, can Grandma end studying me ‘The Wizard of Oz’?”
The Amazon Echo Dot responds with an “Okay” after which speaks in the voice of the boy’s grandmother.
“As you noticed in this expertise, as a substitute of Alexa’s voice studying the e-book, it’s the child’s grandma’s voice,” Prasad instructed attendees.
“We’re unquestionably residing in the golden period of AI, the place our goals and science fiction have gotten a actuality,” he added.
Extra superior than present voice tech
The new voice replication function is in improvement. Neither Prasad nor Amazon has but revealed the date of its public launch.
Amazon already has a voice-synthesizing function that lets Alexa mimic voices of celebrities akin to Shaquille O’Neal and Samuel L. Jackson and even fictional character voices akin to Marvel’s Deadpool and Star Wars’ R2-D2.
However until now the function required dozens of hours of audio recording to have the ability to replicate the voice. The new function, quite the opposite, requires a few seconds of unique voice recording.
Amazon’s newest improvement has triggered a new debate round AI expertise, particularly because it comes shut on the heels of the declare by a Google engineer that the search engine big’s chatbot named LaMDA has gained sentience.
(Important picture: Andres Urena/@andresurena/Unsplash; Featured picture: Nicolas J Leclercq/@nicolasjleclercq/Unsplash)