Amazon is developing a way to memorialize your departed loved ones by incorporating a voice-mimicking feature into Alexa, the tech giant’s voice-controlled virtual assistant.
In Amazon's Re:MARS conference held June 22, senior vice president and Alexa team head scientist Rohit Prasad revealed a new feature that allows Alexa to recreate anybody’s voice, even the deceased.
According to a Reuters’ report, Prasad said that the feature aims to honor those who have passed away amid the pandemic.
“While AI can’t eliminate that pain of loss, it can definitely make the memories last,” said Prasad, adding that many people lost loved ones to COVID.
Focusing on companionship and a more conversational AI (artificial intelligence), TechCrunch said that Amazon developed a system that could mimic audio output after listening to a one-minute speech of a person. This could then be utilized for a longer output similar to the individual’s voice, including relatives who have passed away.
“This required inventions where we had to learn to produce a high-quality voice with less than a minute of recording versus hours of recording in the studio,” the senior vice president noted.
He also said that Amazon framed the problem “as a voice conversion task and not a speech generation path,” mentioning how science fiction is transforming into reality under the golden era of AI.
A spokesperson for Amazon also clarified that the new feature will not be exclusive to replicating voices of the departed but is based on recent text-to-speech developments, as elaborated by the company.
In a demonstration during the conference, Prasad presented a video of a child asking Alexa to finish reading the Wizard of Oz through the voice of the child’s grandmother. The smart assistant responded by switching its default voice to a more humanlike one, nearly identical to that of the relative.
While Amazon is yet to unveil the feature’s release, some netizens on Twitter voiced their concerns about the potential harm of voice-mimicking, citing possible abuse as well as mistrust on the technology.
How this should have gone:— Robert Merkel (@rgmerk) June 22, 2022
Junior engineer: hey, we could use this tech to allow Alexa to imitate people’s voices!
Senior engineer: that could be abused in a million nasty ways. We’re not going to do that.
What actually happened:
Another Twitter user referenced Max Headroom—a major signal hijack dating from 1987 that involved two broadcast stations—and said that those who forgot the controversial incident are doomed to repeat it.
Those who forget Max Headroom are doomed to repeat it. https://t.co/C8GTW9Fg2t— Adam Rogers (@jetjocko) June 23, 2022
Prasad had emphasized that the new feature should not be confused with the “all-knowing, all-capable, uber artificial general intelligence," and only aims for user-environment adaptability and concept-learning with minor external input.
Still, a Twitter user hinted at the eeriness of the feature.
This isn’t creepy at all https://t.co/VE4JOfrHrg— Eric The Red (@RWACPodcast) June 22, 2022
Netizens were also quick to draw comparisons to Black Mirror, a Netflix series about the dark and twisted side of technological innovations.
Have we learned nothing from Black Mirror or Made For Love?🤯Oh, cool: Amazon says Alexa will soon be able to mimic the voices of your beloved dead https://t.co/Ah30DsSXrx— Aldous J Pennyfarthing "Pedal or Perish" (@AJPennyfarthing) June 23, 2022