In a really unusual quick demo, Amazon confirmed how Alexa can mimic the voice of a useless relative to learn bedtime tales or accomplish different duties involving “human-like empathy.” The function remains to be experimental, however based on Amazon, Alexa solely wants a couple of minutes of audio to impersonate somebody.
The demo came about in the course of Amazon’s annual re:MARS convention, an business get-together centered on machine studying, area exploration, and some different highly effective issues. In it, a younger baby asks Alexa if Grandma can learn wizard of oz—The speaker responds accordingly with a composite voice.
“Instead of Alexa’s voice reading the book, it’s the baby’s grandmother’s voice,” Rohit Prasad, Amazon’s chief scientist for Alexa AI, instructed a quiet crowd after the demo.
Prasad notes that “many of us have lost someone we love” to the pandemic, and claims that synthesizing sound utilizing synthetic intelligence can “make their memories last”. That is clearly a controversial concept — it is ethically questionable, we do not know the way it would possibly have an effect on psychological well being, and we’re undecided how far Amazon desires to push the know-how. (I imply, can I take advantage of the voice of a useless relative to navigate the GPS? What is the level right here?)
Amazon’s superior audio synthesis know-how can also be worrisome. Beforehand, Amazon has replicated the voices of celebrities corresponding to Shaquille O’Neal utilizing a number of hours Professionally Recorded Content material. However the firm now claims that it will possibly transcribe an audio with just some minutes of audio. We have already seen how audio synthesis know-how may help with fraud and theft, so what occurs subsequent?
We do not know if Amazon will launch this voice synthesis function on its sensible audio system. However deep voice falsification is basically inevitable. They’re already an enormous a part of the leisure business (see High Gun: Maverick For instance), Amazon is simply one of many many firms attempting to clone voices.
Supply: Amazon through The Verge