Amazon announced at the re:MARS conference that it is working on a feature that will allow it to convert short audio clips of a person's voice into longer speech. The voice of a dead loved one can be used to read a grandson a book.
The way the company made it happen was by framing the problem as a voice conversion task and not a speech generation path
There isn't a lot of information beyond this initial demonstration. The goal of this technology is to make memories last after so many of us have lost someone we love.
The Takara Tomy smart speaker we reported on a few months back was able to mimic a parent's voice at night so they could read to their children even when they weren't around. It is a good idea for parents who travel a lot and are tired of reading the same books. It raises the question of if a fake voice can be as comforting as a real one.
What kind of emotions would be evoked by hearing the voice of a loved one who has died, and what kind of diseases would evoke those emotions? Wouldn't it make them angry when they don't have someone to talk to? How does the interaction with the dead affect someone who is grieving? Does it affect the process or not? What would the dead person think of it?
It is unclear how the feature's voice samples will be stored.
The re:MARS conference is meant to highlight the company's ventures in ambient computing. Machine learning, automation, robots, and space are some of the things referred to as MARS. Over the next few days, it is expected that Amazon will share more information at its Las Vegas conference.