New Amazon Alexa feature can mimic people's voices, including the dead

The new feature was unveiled during the Re:Mars conference held by Amazon in Las Vegas on Wednesday, June 22nd, 2022.

Amazon is currently developing the ability of Alexa to emulate a person's voice while needing less than a minute of recorded audio to do so. This will supposedly imbue Alexa with more "human attributes of empathy and affect," improving interactions with the virtual assistant. The main hurdle to developing the feature arises from having to learn how to create a "high-quality voice" with only snippets of audio, presumably collected by Alexa itself, instead of hours of studio recordings.

"These attributes have become even more important during the ongoing pandemic when so many of us have lost ones that we love. While AI can't eliminate that pain of loss, it can definitely make their memories last," said Rohit Prasad, the senior vice president and head scientist for Alexa, at the event.

Played at the event was a video featuring a young child, in which they ask, "Alexa, can Grandma finish reading me the Wizard of Oz?" Alexa grants the child's request by assuming the voice of the child's grandmother to continue reading the book in her voice. However, while Amazon pushes to develop the feature for Alexa, Microsoft is proceeding more cautiously regarding the voice emulation technology.

"This technology has exciting potential in education, accessibility, and entertainment, and yet it is also easy to imagine how it could be used to inappropriately impersonate speakers and deceive listeners," said Natasha Crampton, who heads Microsoft's AI ethics division, in a blog post.