Amazon’s Alexa may soon be able to reproduce the voices of family members – even if they’re dead.
The capability, announced at Amazon’s Re:Mars conference in Las Vegas, is under development and will allow the virtual assistant to mimic a specific person’s voice based on an audio recording provided with less than one hour of audio. minute.
Rohit Prasad, Alexa senior vice president and chief scientist, said at Wednesday’s event that the desire behind the feature was to build greater trust in the interactions users have with Alexa using Alexa. adding “human attributes of empathy and influence”.
“These attributes become even more important during this ongoing pandemic when so many of us have lost the things we hold dear,” says Prasad. “While AI can’t erase that pain of loss, it can certainly make their memories last.”
In a video broadcast by Amazon at the event, a child asks “Alexa, can you read me The Wizard of Oz?” Alexa then acknowledges the request and switches to another voice that mimics the child’s grandmother. The voice assistant then continues to read the book with that voice.
To make this feature available, Prasad said the company had to learn to create “high-quality voices” with shorter recordings, rather than recording hours in the studio.
Amazon did not provide further details about the feature. Certain implementations will raise many privacy concerns and ethical questions about consent.