As bizarre as this may sound, this is not science fiction. After community projects to emulate some iconic voices such as Morgan Freeman’s, this time Amazon itself has announced that it is working on a new artificial intelligence project for Alexa customization, with which we can emulate the voice of any person.
This bizarre revelation was made at the company’s annual re:Mars conference, where Amazon Senior Vice President and Alexa Chief Scientist Rohit Prasad detailed a number of new features coming to the smart assistant in the future. However, the most striking announcement was without a doubt this small demo in which they showed Alexa reading a story to a child before going to sleep, using the voice of his deceased grandmother.
A miracle that was achieved with just one minute of audio of the woman, just as Prasad himself explained during the conference, «this required inventions in which we had to learn to produce a high-quality voice with less than a minute of recording time versus hours of recording in the studio. The way we made it happen was by framing the problem as a speech conversion task and not a speech generation path.«.
However, at the moment it seems that it is just a demonstration of its capabilities rather than a functionality that we are going to see soon available for the voice assistant. And it is that Amazon not only has not shared any details about its possible date, but also avoided talking about general availability for the moment. However, it is clear that this functionality has great appeal, so it remains to be seen how Amazon will approach its commercialization.
Having said that, We will have to see the limitations established by the company for this function, since it could lead to some problems in the face of identity theft. Even so, it is certainly inspiring to see the great advances that artificial intelligence continues to offer, overcoming barriers that we practically considered unattainable to propose a future full of possibilities.
