Alexa will be able to mimic individuals after hearing an audio recording, the company says
Amazon has unveiled plans to develop a system that would allow its Alexa voice assistant to mimic people's voices after hearing a short audio recording. The new feature was announced during the company’s re:MARS conference in Las Vegas on Wednesday.
Amazon’s senior vice president and head scientist of Alexa, Rohit Prasad, showcased the new feature for attendees of the conference, and explained that the aim of this system is to “make the memories last.”
During the demonstration of the feature, a child asked the voice assistant if “Grandma can finish reading me the Wizard of Oz?” After a few moments, Alexa confirmed the command and apparently began speaking in the voice of the child’s grandmother.
Prasad stated that this new feature, which is still in development and has no announced release date, 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.”
He explained that the way they made it happen was by framing the problem as “a voice conversion task” and not a “speech generation path,” noting that “we are unquestionably living in the golden era of artificial intelligence, where our dreams and science fictions are becoming a reality.”
The Amazon VP suggested that, while the feature could ostensibly be used to replicate any voice, it could also be used to help memorialize a deceased family member. “So many of us have lost someone we love” during the Covid-19 pandemic, said Prasad, noting that making artificial intelligence conversational and companion-like has become a key focus of the team.
“While AI can’t eliminate that pain of loss, it can definitely make the memories last,” he added.
Amazon has said it wants to make its Alexa voice assistant feel more natural overall, and has been consistently improving the functionality of the program. The e-commerce giant has been rolling out features that enable it to hold more human-like conversations with consumers, even to the point of asking questions.
