Alexa could soon have a ‘creepy’ new ability to mimic dead voices

Smart speaker stock photo.
Smart speaker stock photo. Photo credit Getty Images

During Amazon’s re:MARS conference in Las Vegas, Nev., this week, Rohit Prasad – senior vice president and head scientist for Alexashowed a video of the artificial intelligence voice service speaking in the voice of a young boy’s grandmother.

“We had to learn to produce a high-quality voice with less than a minute of recording,” Prasad said of Alexa’s new ability. In the example, the boy asked his grandmother to finish reading him “The Wizard of Oz”.
Prasad explained that developers were able to add the feature as a “voice conversion” task rather than a “speech generation” task.

While Prasad went on to say that “we are unquestionably living in the golden era of AI, where our dreams and science fiction are becoming a reality,” some people have had less than positive reactions to the feature.

Andy Geirsher of WBBM’s Noon Business Hour noted that people already do things such as save voicemails or videos of their loved ones, but he questioned whether they would want to take things further.

“Do they also want that long departed relative reading them the weather or giving them directions in their car?” he asked Jennifer Jolly, a Tech Life Columnist for USA Today and Editor-in-Chief of Techish.

“There are so many issues to talk about with this type of technology,” she said, adding that new feature was “creepy” and “just weird.”

Geirsher said that similar technology was used in a positive way by Roger Ebert, the late Chicago-area film critic, when he was unable to speak due to medical treatments. On the other hand, he said that many found Kanye West’s use of a hologram of his ex-wife Kim Kardashian’s late father Robert Kardashian at her 2020 birthday party off-putting.

Jolly said she has concerns over using the voice of people who have passed on – and therefore cannot consent to their voice being used – and the use of children’s voices.

“This is one of those kinds of metaverse issues where, who gets to decide what’s kind of morally okay,” she said.

Already, artificial intelligence technology that can alter images has been used to create “deepfake” videos, such as ones posted of Tom Cruise impersonator Miles Fischer on TikTok, per a CNN report.

Jolly also stressed that Amazon did not say when, or even if, they will offer the new voice feature to customers. She said that the security concerns around being able to recreate a voice make her skeptical of the company “turning that loose to the general public.”

“It’s a good time to sort of talk about where our boundaries are,” Jolly said. “I, as a consumer technology reporter, hope that I don’t see this anytime soon.”

Follow KNX News 97.1 FM
Twitter | Facebook | Instagram | TikTok