At the annual in-house re:Mars conference for machine learning, automation and robotics in Las Vegas, Amazon's senior vice president and head scientist for Alexa, Rohit Prasad, presented new functions for the digital The company's language assistant, Alexa, promised – for example, to adopt the voice of the dead grandmother.
A short recording is enough for voice-making
< p class="p text-width">Following the direction of the conference, he demonstrated a potentially new function in which a short audio excerpt serves as the basis for a longer, synthesized speech output. Unlike the official voices of the digital language assistant, for which the speakers are recorded for hours in the studio, a one-minute audio clip is enough to tell a story with a new voice.
Deceased grandma reads a story to grandchildren
At the conference, Amazon showcased improvements in voice creation technologies by taking a one-minute audio recording of a deceased grandmother and then reading a grandchild a bedtime story in that voice—in this case, “The The wizard of Oz”. Rohit Prasad described the new possibilities as saying that “we are undeniably living in the golden era of artificial intelligence, where dreams and science fiction come true”.
According to Prasad the function is meant to honor those who have died during the pandemic and while AI cannot take mourning, it can preserve memories.
The keynote of Rohit Prasad's lecture on the right :Mars is available on YouTube and from 1:01:58 the new feature will be demonstrated.
Market launch unknown
It is questionable whether Amazon will ever actually introduce this function for the general public, further details on the plans are not yet available. Aside from the function shown, of having a story read aloud with the voice of a deceased relative, there are also many other possibilities that are likely to cause discussion and skepticism.