At its annual re:Mars convention at present in Las Vegas, Amazon’s Senior Vice President and Head Scientist for Alexa, Rohit Prasad, introduced a spate of recent and upcoming options for the corporate’s sensible assistant. Probably the most head turning of the bunch was a possible new characteristic that may synthesize quick audio clips into longer speech.
Within the situation introduced on the occasion, the voice of a deceased beloved one (a grandmother, on this case), is used to learn a grandson a bedtime story. Prasad notes that, utilizing the brand new expertise, the corporate is ready to accomplish some very spectacular audio output, utilizing only one minute of speech.
“This required innovations the place we needed to study to provide a high-quality voice with lower than a minute of recording versus hours of recording the studio,” the manager notes. “The best way we made it occur is by framing the issue as a voice conversion job and never a speech technology path. We’re on questionably, dwelling within the golden period of AI, the place our goals and science.”
Particulars are scant, in the meanwhile. There’s no timeline or additional specifics, however – at very least – that is the type of information that can possible invite all method of scrutiny over potential purposes past one thing as banal and even heartwarming as studying a toddler The Wizard of Oz.
