Press "Enter" to skip to content

Alexa will be able to Mimic Deceased Loved Ones’ Voices

The most recent version of Amazon’s speech assistant Alexa could seem uncannily similar. The business revealed that it is developing an upgrade to its Alexa system that would enable the technology to imitate any voice, including that of a departed family member, on Wednesday during its annual re:MARS conference, which focuses on artificial intelligence research.

In a video played on stage, the E-commerce giant showed how the grandma of a small child was reading a story to him instead of Alexa trademark voice. The improved technology, according to Rohit Prasad, a senior vice president at Amazon, would be able to gather enough speech data from less than a minute of audio to enable customisation like this, as opposed to requiring someone to spend hours in a recording studio as it has in the past. When this functionality could go live was not specified by Prasad. Regarding a timetable, Amazon declined to respond.

Amazon came up with the idea after considering fresh approaches to give artificial intelligence(AI) more “human traits,” particularly “in these times of the global epidemic, when so many of us have lost someone we love,” according to Prasad. “AI can surely help their memories linger, even though it can’t take away the agony of loss.”

Amazon has long employed well-known voices to voice Alexa, including the actual voices of Samuel L. Jackson, Melissa McCarthy, and Shaquille O’Neal. But over the past few years, AI speech recreations have also become better and better, especially with the aid of deepfake technology. For instance, three statements in the Anthony Bourdain documentary “Roadrunner” that appeared to have come from the late media personality were really produced by AI. When the movie appeared last year, director Morgan Neville told The New Yorker, “We can have a documentary-ethics panel about it afterwards.”

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *