According to an Engadget post, Amazon’s Alexa has a new skillset. Alexa’s senior vice-president Rohit Prasad exhibited Alexa’s ability to mimic voices at Amazon’s Re:Mars conference. The demonstration included Alexa “reading to a child in a voice of his recently deceased grandmother.” All of the major tech companies are using some form of AI in products including Amazon. AI is a big driver in how Alexa is able to simulate a voice. As Prasad stated, “While AI can’t eliminate that pain of loss, it can definitely make the memories last.”
Supposedly, Alexa can create a credible voice clone with as little as one minute of sample source audio. We’ve been concerned about deep fake audios for some time. The latest Alexa demo seems to put a consumer marketing spin on the capabilities. If it’s this easy to create a deep fake audio, there’s going to be all kinds of unintended consequences including new forms of scams. The criminals just got a new tool for their toolbox.