Amazon has announced that it is seeking to develop a new feature for voice assistant Alexa that will help it mimic the voice of anyone a user may want it to sound like. In its demonstration, the company showed the example of a grandmother’s voice.
The announcement was made by Amazon Senior Vice President Rohan Prasad at a company conference in Las Vegas, according to Reuters.
According to Prasad, the goal of this new feature is to “make memories last”. In specific, he mentioned that the company believes that this feature could be a soothing tool for those who “have lost someone we love” during the pandemic.
Although Prasad did not mention a timeline for the launch of this new feature, he did mention that it would need less than minute’s audio to mimic a voice.
In Amazon’s demonstration, it showed a child asking Alexa for his grandmother to read him a story. Alexa then morphed its voice to a less robotic and more soothing one.
There have been concerns that tools such as this could be used for political deepfakes and other dangerous propaganda,
Prasad clarified that they are not looking to build artificial general intelligence (AGI) into their products as compared to competitors like Google’s DeepMind project and Elon Musk’s OpenAI.
